Abstract: Today, all microprocessor-based measurement and control systems require sensors to provide data for real-time decision-making. This paper details the hardware design of a pressure sensor measurement system, the intelligent software design of the pressure sensor, and the design of pressure sensor error and temperature compensation techniques. To address issues such as temperature drift and nonlinearity in sensor measurements, a solution is proposed using multi-sensor information fusion techniques—surface fitting and curve fitting. Communication between the PC and the sensor measurement system is implemented to perform data conversion, data processing, and printing functions, thus improving the measurement system. Keywords: Intelligent pressure sensor; PC; Temperature compensation [b][align=center]Design of Pressure Measurement System based on Multi-sensor Technology WU Zhi[/align][/b] Abstract: At present, all control systems based on microprocessors need sensors to supply real-time data. The author detailed the hardware design of the measurement system and the software design of the intelligent sensor. We proposed a scheme that applied multi-sensor data fusion-surface fitting and curve-fitting to this system based on the temperature drift and nonlinearity of the sensor. Communication between the personal computer and the sensor measurement system was realized by a transceiver, which could complete the data conversion, data processing, and printing tasks, resulting in a further improvement of the measurement system. Keywords: Intelligent Pressure Sensor; PC; Temperature Compensation 1 Introduction The research on sensors began in the 1930s. It is an interdisciplinary and marginal technology science that studies the conversion between non-electrical information and electrical information. Early sensors were analog sensors, now often referred to as traditional sensors. With the development of high-performance computer measurement and control systems, traditional sensors are no longer suitable for systems that require digitalization and intelligence. Control systems require sensors to output digital signals and possess strong information processing and self-management capabilities to achieve information acquisition and preprocessing, reduce the data processing burden on the control computer, and improve the reliability of the entire measurement and control system. 2. Intelligent Pressure Sensor Hardware Design Figure 1 shows the overall system structure diagram. As can be seen from the diagram, the hardware circuit is simple. The ADμC812 microcontroller does not require external A/D and D/A converters, does not occupy a large amount of space, and has a reprogrammable non-volatile flash/electrically erased program memory, making debugging simple. [align=center] Figure 1 Overall System Structure Diagram[/align] 2.1 Signal Analysis and Processing Previous hardware designs used multiplexers to control channel selection, then used sample-and-hold circuits and AID converters for analog-to-digital conversion, and finally sent the converted digital signal to the microcontroller. These discrete components introduced many errors and occupied a large amount of space. The ADμC812 microcontroller used in this paper is a truly complete data acquisition system chip (including data conversion circuitry, microcontroller, flash/electrically erasable memory, etc.). The microconverter integrates a fully programmable, high-precision analog data acquisition system. It is small enough to be placed within a sensor, transponder, or cable connector. Its extremely low cost allows it to replace high-cost, bulky products based on single-board computers. Due to its high precision and speed, it is suitable for intelligent sensing, instantaneous acquisition, data acquisition, and various communication systems. Its functional block diagram is shown in Figure 2. [align=center]Figure 2 Functional Block Diagram of ADμC812[/align] In this design, P1.0, P1.1, P1.2, and P1.3 of the ADμC812 microcontroller's P1 port are used as four signal input channels. One channel inputs differential pressure signal, one inputs temperature signal, one inputs static pressure signal, and one is grounded. This grounded channel can be used with appropriate software to reduce temperature drift and system errors. P1.7 is connected to an LED to monitor whether the microcontroller is working properly. P2.0 and P2.1 of the P2 port are used as input pulses and data terminals for the LCD display. P3.0 (RXD) and P3.1 (TXD) of the P3 port are connected to an external MAX232 for level conversion to achieve communication with the PC. This design uses an external clock generation method with a crystal frequency of 11.0592MHz and an internal reference. A 0.1μF capacitor is connected between pin 7 (CRER) and AGND. The power reset circuit uses a MAX708 chip for reset. 2.2 Sensor Output Signal Preprocessing The preprocessing of the sensor output signal mainly consists of a signal conditioning circuit, including the design of constant voltage source circuits for differential pressure, static pressure, and temperature sensors, as well as filtering and amplifying the sensor output signal. 1. Constant Voltage Source Circuit Design. Without voltage regulation, the voltage will spike when the sensor is initially powered, causing the 5V supplied to the sensor to exceed 5V. This can damage the intelligent sensor and, in severe cases, burn out the entire circuit. Therefore, this paper uses secondary voltage regulation to protect the intelligent pressure sensor. The voltage regulator used in this paper is the LM117, which has only three pins: input, output, and common. It requires few external components and is convenient and reliable to use. During measurement, the time drift and temperature drift of the sensor are affected by the environment. To improve its stability, this paper selects the REF192 reference voltage source for secondary voltage regulation of the pressure, static pressure, and temperature sensors. The REF192 is suitable for powering intelligent sensors and has the advantages of small temperature drift, low output noise, low dynamic internal resistance, and short-circuit protection compared with ordinary reference power supplies. The schematic diagram is shown in Figure 3. The main parameters of the REF192 are as follows: reference voltage VR = 2.5V, temperature coefficient TC < 5PPm/℃, and output noise voltage of 25μV. When the reference voltage VR = 2.5V and the temperature range is 10℃～80℃, the output voltage of the temperature sensor varies from 1.5V to 2V. 2. Figure 3 is the schematic diagram for signal amplification. The amplifier uses the AD623 single-supply instrumentation amplifier, which is characterized by ease of use, low power consumption, high accuracy, and low noise. The AD623 uses a setting resistor for gain control, which is determined by the impedance between pin 1 and pin 8. The pressure signal variation range is 0mV to 500mV, and the amplification factor is set to 5 times. The electrical signal after sensor conversion and amplification contains noise signals of various frequencies due to various reasons. Filtering measures are needed to suppress unwanted stray signals. We construct a low-pass filter using R3 and C3 to increase the system's signal-to-noise ratio. Here, C3 = 0.1μF, R3 = 1K. [align=center] Figure 3 Signal Amplification Circuit[/align] 3 Intelligent Pressure Sensor Software Design The software of the intelligent pressure sensor consists of a monitoring program, an interrupt program, a measurement program, and a signal processing program. Under the management of the main program, the system can self-check whether the RAM and A/D conversion chip are functioning properly. If a fault occurs, it can display which component is malfunctioning and the detailed location of the error, so that the operator can handle it promptly; if normal, it initializes the system, including setting the stack pointer, initializing the timer/counter, initializing the interrupt, and initializing relevant working units. 1. Monitoring Program Design. The monitoring program is the main thread in the intelligent sensor software. It calls various modules and connects them to form an organic whole, realizing the full management functions of the instrument. In this design, since the CPU used does not have ROM and E2PROM, software design is saved. We only perform self-tests on RAM and A/D. Since the CPU used in this paper automatically performs a self-test on A/D when powered on, its self-test program does not need to be designed. 2. Measurement control program. The measurement control program completes the measurement and control tasks of the measurement process, such as multi-channel switching, sampling, A/D conversion, etc. These functions can be implemented by several program modules, which can be called by the monitoring program or interrupt program. The measurement program needs to acquire temperature, static pressure, and differential pressure signals, perform linear correction on the differential pressure, and then compensate the differential pressure signal with temperature and static pressure signals. After compensation, the differential pressure measurement value is displayed. 3. Data communication software design. We use LEDs to display data in real time. Since LEDs can only display data and cannot analyze it, communication between the system and the PC is realized. The PC can be used to display, process, and print the measurement results. This part of the software includes the communication software design for the microcontroller and the communication software design for the PC. The PC communication software design is completed using C++Builder. The design of the PC communication software is divided into four steps: 1) Opening and initializing the serial port. Initialization involves opening the port in a specific way and setting the baud rate, number of data bits, number of stop bits, parity check mode, error detection, timeout parameters, and response events; 2) Sending data, writing the data to be sent to the port and handling errors; 3) Receiving data, reading the port data after the port receives an event character and places it in the input buffer; 4) Closing the port and releasing system resources. The communication software design for the intelligent sensor is also included. This paper uses half-duplex mode to transmit data. Microcontroller serial communication has four working modes; we use mode 1, which is characterized by a variable baud rate and no parity bit. The baud rate setting is related to the overflow rate of Timer 1. This paper uses mode 2 of the four operating modes of timer T1. In mode 2, TL0 is used as an 8-bit counter, and TH0 is used to store the initial value. Whenever TL0 overflows, the initial value in TH0 is automatically reloaded into TH0, and this process repeats. The baud rate formula is: Baud rate = 2[sup]SMOD[/sup] × crystal frequency / [32 × 12 × (256 - X)]. The baud rate used in this paper is 9600, the crystal frequency is 11.0592, SMOD is set to 0, and TH1 is 253. The block diagram of the microcontroller communication subroutine is shown in Figure 4. [align=center]Figure 4. Block Diagram of Microcontroller Communication Subroutine[/align] 4. Error and Temperature Compensation Technology Design Any measurement process inevitably involves errors. Based on the pattern of error occurrence, these errors can be divided into random errors and systematic errors. Random errors can be overcome using filtering methods, while systematic errors cannot be eliminated by statistical averaging. Instead, specific measures must be taken in measurement technology for a particular situation. This paper focuses on minimizing these two types of errors in intelligent sensors and improving the accuracy of collected data. We have designed corresponding software to complement the hardware structure. 1. Random Errors. Random interference introduces useless components into the measured signal. Filters can be used to remove these useless components and improve signal quality. Commonly used digital filtering algorithms include amplitude limiting filtering, median filtering, arithmetic mean filtering, recursive average filtering, first-order inertial filtering, and composite filtering. Since composite filtering can remove pulse interference and smooth the sampled values, combining the advantages of median filtering and arithmetic mean filtering, this paper adopts the composite filtering algorithm. The algorithm principle of composite filtering is as follows: First, the N sampled values ​​are sorted by size, then the maximum and minimum values ​​are removed, and then the arithmetic mean of the remaining N-2 sampled values ​​is calculated. 2. System error. (1) Automatic zero-point calibration. In order to eliminate the instrument zero-point drift caused by changes in the output of the sensor or the gain of the amplifier due to changes in environmental factors, and thus cause system error, we adopt the method of automatic zero-point calibration. The CPU used in this paper has 8 input channels, so in addition to the temperature, static pressure and differential pressure input channels, another channel is grounded. In this way, when collecting data, the value obtained by the grounded channel can be directly subtracted by software technology. (2) Nonlinear correction. In order to realize that the input-output characteristic is a straight line, that is, the sensitivity is a constant within the measurement range, this paper adopts the curve fitting method and sets an nth degree polynomial of Taylor expansion to approximate the nonlinear curve of sensor measurement. (3) Temperature compensation. This paper mainly adopts the multidimensional regression analysis method in the information fusion processing method. Its basic idea is: to use the multidimensional regression equation to establish the correspondence between the measured target parameter and the sensor output. Unlike classic one-dimensional sensor calibration experiments, this method involves multi-dimensional calibration experiments. Then, using the least squares method, the coefficients in the regression equation under the condition of minimum mean square error are calculated from the experimental calibration data. Thus, when the sensor's output value is measured, the corresponding input target parameters can be calculated from the multi-dimensional regression equation with known coefficients. Nonlinear correction and temperature compensation require several calls to floating-point addition and multiplication subroutines to obtain the results. The author's innovation lies in: This paper uses the ADμC812 microcontroller to design the hardware circuit. This chip integrates a high-performance 8-bit MCU with reprogrammable non-volatile flash/electrically erased program memory, as well as a high-performance self-calibrating 8-channel ADC and a 2-channel 12-bit DAC, simplifying the hardware circuit design, reducing its size, portability, and error. Communication with a PC is used to complete data conversion, data processing, and data printing functions, facilitating centralized system monitoring. In terms of anti-interference, this paper uses a low-pass filter in hardware and a composite filtering method in software to overcome random errors, thereby improving the stability of the system in different situations. References: [1] Tian Liqiang, Sun Yicai, Zhao Yanxiao. Comparison of two multi-sensor information fusion methods [J]. Sensor World. 2002. VOL1, 12: 9-13 [2] Wang Huiying, Ma Xuetong, Wang Lei. Implementation of asynchronous serial communication under Windows 9x environment based on ActiveX control [J]. Computer-Aided Engineering. 2002. VOL12, 4: 65-69 [3] Liu Qingsong, Qian Suxiang, Yan Gongbiao. Application of information data fusion technology based on multi-sensor in power system [J]. Microcomputer Information, 2006, 11-1: 191-193