Intelligent pressure sensor temperature compensation system

Abstract

The invention provides an intelligent pressure sensor temperature compensation system, and relates to the technical field of pressure sensors, and the system comprises a hardware sensing module which comprises an elastic sensitive element and is used for reducing the temperature drift magnitude of a pressure sensor; the environment data acquisition module comprises a flexible thermoelectric sensor array and is used for acquiring temperature and humidity data; the humidity correction module is used for integrating a dynamic humidity correction algorithm and establishing a cross-sensitive matrix model to correct the original signal to generate a first pressure correction signal; the temperature compensation module is used for carrying an incremental neural compensation network and constructing a compensation model to perform temperature compensation on the first pressure correction signal to generate a second pressure compensation signal; and the fault compensation module is used for triggering a digital twinborn compensation model to perform fault compensation on the second pressure compensation signal to generate a target pressure output signal when a fault of the single sensor is monitored, so that the pressure measurement error in the whole temperature range can be accurately controlled, and the environmental adaptability and the intelligent level of the system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of pressure sensors, and in particular to an intelligent pressure sensor temperature compensation system. Background Art

[0002] In many fields such as industrial automation, aerospace, etc., pressure sensors are key components for achieving precise measurement and control.

[0003] However, in practical applications, temperature changes can have a significant impact on the measurement accuracy of pressure sensors, mainly manifesting as temperature drift. Traditional pressure sensor temperature compensation methods often only compensate for a single temperature point or a simple temperature change trend, making it difficult to accurately capture the spatial gradient changes in the temperature field in complex environments. At the same time, the cross-sensitivity effect between humidity and temperature is often ignored, resulting in a significant increase in measurement errors in high humidity or environments with drastic humidity changes. In addition, most existing compensation algorithms use fixed parameter models, which are difficult to adapt to dynamic working conditions. They have shortcomings in real-time performance and model adaptability, and cannot meet the high requirements for precision measurement in harsh environments such as aerospace and industrial automation.

[0004] Therefore, it is necessary to provide an intelligent pressure sensor temperature compensation system to solve the above technical problems. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides an intelligent pressure sensor temperature compensation system, which is used to solve the problems that the existing technology cannot accurately control the pressure measurement error in the entire temperature range, has poor environmental adaptability and real-time response, and has a low level of intelligence.

[0006] The present invention provides an intelligent pressure sensor temperature compensation system, the system comprising: A hardware sensing module, including an elastic sensitive element made of a double negative temperature coefficient composite material, is used to reduce the original temperature drift magnitude of the pressure sensor; An environmental data acquisition module, comprising a flexible thermoelectric sensor array, arranged on the surface of the pressure sensor, for acquiring temperature spatial distribution data and dynamic humidity data of the pressure sensor; a humidity correction module, configured to integrate a dynamic humidity correction algorithm, establish a cross-sensitivity matrix model of humidity and pressure, correct the original pressure output signal of the pressure sensor, and generate a first pressure correction signal; a temperature compensation module, configured to carry an incremental neural compensation network based on transfer learning and build a millisecond-level temperature compensation model based on an embedded edge computing device, and perform temperature compensation on the first pressure correction signal to generate a second pressure compensation signal; A fault compensation module is used to automatically trigger the digital twin compensation model to perform fault compensation on the second pressure compensation signal and generate a target pressure output signal when a single sensor node fault in the flexible thermoelectric sensor array is detected.

[0007] Preferably, the thermal expansion coefficient of the elastic sensitive element presents a two-stage negative temperature response performance in the temperature range of -40°C to 125°C, and is less than or equal to 15×10⁻ at 25°C. 6 / ℃.

[0008] Preferably, the double negative temperature coefficient composite material includes a main material and a negative thermal expansion filler; the main material is a polyetheretherketone-based polymer material, the corresponding volume proportion is 60% to 70%, and the melt index is less than or equal to 15g / 10min; the negative thermal expansion filler is a composite particle of tungsten carbide and carbon fiber, the corresponding particle size is 5μm to 10μm, the corresponding carbon fiber aspect ratio is greater than or equal to 100, and the corresponding volume proportion is 30% to 40%; After the double negative temperature coefficient composite material is compression molded, the temperature coefficient of elastic modulus at 25° C. is less than or equal to -800 MPa / ° C.

[0009] Preferably, the flexible thermoelectric sensor array includes a polyimide flexible substrate and grid-shaped distributed nodes; the thickness of the polyimide flexible substrate is 5μm to 15μm, and a micro-electromechanical thermopile sensor is integrated on the surface of the polyimide flexible substrate, and the response time of a single micro-electromechanical thermopile sensor is less than or equal to 50ms; the grid-shaped distributed nodes are integrated with a temperature and voltage conversion unit and a 24-bit analog-to-digital conversion module, support temperature field spatial sampling with a resolution of 0.1°C, and the sampling frequency is greater than or equal to 1000Hz.

[0010] Preferably, the elastic sensitive element and the flexible thermoelectric sensor array are integrated through an integrated packaging process, and the packaging material is an organic silicone rubber with a thermal conductivity greater than or equal to 1.5 W / (m·K).

[0011] Preferably, the humidity correction module is used to integrate a dynamic humidity correction algorithm, establish a cross-sensitivity matrix model of humidity and pressure, correct the original pressure output signal of the pressure sensor, and generate a first pressure correction signal, specifically including: a relationship matrix establishing unit, configured to divide the dynamic humidity data into a plurality of humidity intervals based on the dynamic humidity correction algorithm, and establish a mapping relationship matrix between the humidity value in each humidity interval and the zero point drift and sensitivity drift of the original pressure output signal through an orthogonal experimental method; a humidity sequence generating unit, configured to dynamically fit the cross-sensitivity matrix model of humidity and pressure based on the mapping relationship matrix using a least squares method and generate a humidity correction coefficient sequence; The pressure signal correction unit is used to correct the original pressure output signal based on the humidity correction coefficient sequence to generate the first pressure correction signal, and at the same time introduce a first-order inertial filter to eliminate humidity mutation noise in the first pressure correction signal.

[0012] Preferably, the temperature compensation module is configured to carry an incremental neural compensation network based on transfer learning, and to construct a millisecond-level temperature compensation model based on an embedded edge computing device, to perform temperature compensation on the first pressure correction signal to generate a second pressure compensation signal, specifically comprising: a temperature matrix generating unit, configured to calculate the spatial gradient of the temperature field based on the temperature spatial distribution data and construct a three-dimensional temperature distribution matrix of the pressure sensor, and introduce the three-dimensional temperature distribution matrix into the incremental neural compensation network based on transfer learning; a signal temperature compensation unit, configured to construct the millisecond-level temperature compensation model based on the embedded edge computing device, and in combination with the incremental neural compensation network based on transfer learning, perform temperature compensation on the first pressure correction signal to generate the second pressure compensation signal and upload the signal to the cloud server corresponding to the pressure sensor; The temperature compensation model training unit is used to aggregate the second pressure compensation signal through the cloud server using a federated learning algorithm and construct an adversarial sample to train the generalization capability of the millisecond-level temperature compensation model.

[0013] Preferably, the fault compensation module is used to automatically trigger the digital twin compensation model to perform fault compensation on the second pressure compensation signal and generate a target pressure output signal when a single sensor node fault in the flexible thermoelectric sensor array is detected, specifically including: a faulty node determining unit, configured to determine a faulty single sensor node by comparing temperature gradient differences between adjacent single sensor nodes in the flexible thermoelectric sensor array and identifying abnormal temperature gradient differences using a Dickson test method; The target signal generation unit is used to automatically trigger the digital twin compensation model to perform fault compensation on the second pressure compensation signal after determining that the single sensor node has a fault, and generate the target pressure output signal, and the digital twin compensation model includes a sensor physical model based on finite element simulation and an operation status prediction model based on LSTM.

[0014] An intelligent pressure sensor temperature compensation method, the method comprising: Reduce the original temperature drift magnitude of the pressure sensor by using elastic sensitive elements; Collecting temperature spatial distribution data and dynamic humidity data of the pressure sensor through a flexible thermoelectric sensor array; Based on a dynamic humidity correction algorithm, a cross-sensitivity matrix model of humidity and pressure is established to correct the original pressure output signal of the pressure sensor to generate a first pressure correction signal; By using an incremental neural compensation network based on transfer learning and building a millisecond-level temperature compensation model based on an embedded edge computing device, the first pressure correction signal is temperature compensated to generate a second pressure compensation signal; When a single sensor node failure in the flexible thermoelectric sensor array is detected, the digital twin compensation model is automatically triggered to perform fault compensation on the second pressure compensation signal to generate a target pressure output signal.

[0015] Compared with related technologies, the intelligent pressure sensor temperature compensation system provided by the present invention has the following beneficial effects: The present invention reduces the original temperature drift magnitude of the pressure sensor through an elastic sensitive element; collects the temperature spatial distribution data and dynamic humidity data of the pressure sensor through a flexible thermoelectric sensor array; establishes a cross-sensitive matrix model of humidity and pressure based on a dynamic humidity correction algorithm, corrects the original pressure output signal of the pressure sensor, and generates a first pressure correction signal; constructs a millisecond-level temperature compensation model based on an embedded edge computing device through an incremental neural compensation network based on transfer learning, and performs temperature compensation on the first pressure correction signal to generate a second pressure compensation signal; when a single sensor node failure is detected in the flexible thermoelectric sensor array, the digital twin compensation model is automatically triggered to compensate for the fault of the second pressure compensation signal and generate a target pressure output signal, thereby achieving precise control of pressure measurement errors in the entire temperature range and improving the adaptability and intelligence level of the system in harsh environments.

[0016] The present invention can physically suppress temperature drift through the elastic sensitive element of the double negative temperature coefficient composite material, reduce the sensitivity of the sensor to ambient temperature changes, lay the hardware foundation for high-precision measurement, and capture the spatial gradient of the temperature field through a flexible thermoelectric sensor array, breaking through the limitations of traditional single-point temperature measurement, enabling the system to accurately perceive the temperature distribution characteristics in complex environments, and providing comprehensive environmental data support for the compensation algorithm. The system of the present invention forms an intelligent processing mechanism of first correction and then compensation through a dynamic humidity correction algorithm and an incremental neural compensation network based on transfer learning, effectively eliminating the cross-sensitivity effect of humidity and temperature. At the same time, it achieves real-time response through embedded edge computing and relies on the cloud platform to achieve continuous model evolution, so that the system has the ability to dynamically adapt to different working conditions. By integrating the digital twin compensation mode, the present invention can seamlessly switch when a single sensor fails, ensuring the stable operation of the system under abnormal conditions. The system of the present invention constructs a multi-level compensation system to achieve precise control of pressure measurement errors in the entire temperature range, significantly improving the adaptability and reliability of the sensor in harsh environments such as aerospace and industrial automation, and providing an efficient solution for precision measurement scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a system block diagram of an intelligent pressure sensor temperature compensation system of the present invention; Figure 2 The figure is a flow chart of an intelligent pressure sensor temperature compensation method of the present invention. DETAILED DESCRIPTION

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0019] Example 1 like Figure 1 As shown, an intelligent pressure sensor temperature compensation system, the system includes: A hardware sensing module, including an elastic sensitive element made of a double negative temperature coefficient composite material, is used to reduce the original temperature drift magnitude of the pressure sensor; The intelligent pressure sensor temperature compensation system of the present invention uses multi-module collaborative design to build a full-chain precision measurement system covering hardware perception, environmental data acquisition, humidity correction, temperature compensation and fault tolerance.

[0020] The hardware sensing module, centered around an elastic sensor made of a dual-negative temperature coefficient composite material, suppresses temperature drift by manipulating the thermal properties of the material. The composite material is manufactured using a polymer base material and a negative thermal expansion filler. By leveraging the negative thermal expansion coefficient of the material, the material's deformation during temperature changes offsets the elastic modulus drift of the pressure sensor caused by thermal effects, essentially reducing the raw temperature error. The raw temperature drift magnitude refers to the measurement error caused by ambient temperature changes when the pressure sensor is not performing any temperature compensation.

[0021] As a key sensing component of the pressure sensor, the elastic sensitive element's thermomechanical properties directly determine the pressure sensor's basic tolerance to ambient temperature fluctuations, providing a low-drift hardware benchmark for subsequent compensation algorithms.

[0022] An environmental data acquisition module, comprising a flexible thermoelectric sensor array, arranged on the surface of the pressure sensor, for acquiring temperature spatial distribution data and dynamic humidity data of the pressure sensor; The environmental data acquisition module can be based on a distributed sensor network built on a flexible substrate, and can achieve real-time perception of multi-dimensional environmental parameters through a high-density thermoelectric sensor array.

[0023] This flexible thermoelectric sensor array utilizes micro-thermoelectric conversion units integrated on a flexible substrate, adhered to the surface of the pressure sensor in a grid-like topology, to form a temperature field monitoring network with spatial resolution. Its core function is to break through the limitations of traditional single-point temperature measurement by capturing the spatial gradient characteristics of the temperature distribution in the measured area. Simultaneously, it collects dynamic humidity parameters, providing multidimensional environmental data input that incorporates the temporal and spatial variations of temperature and the coupled effects of humidity for subsequent compensation algorithms, thus solving the measurement problem of non-uniform distribution of environmental parameters under complex working conditions.

[0024] a humidity correction module, configured to integrate a dynamic humidity correction algorithm, establish a cross-sensitivity matrix model of humidity and pressure, correct the original pressure output signal of the pressure sensor, and generate a first pressure correction signal; The dynamic humidity correction algorithm is an adaptive correction method for the cross-sensitivity effect between humidity and pressure. The humidity-pressure cross-sensitivity matrix model is a mathematical model that describes the coupled relationship between humidity and pressure. It represents the quantitative relationship between sensor zero drift and sensitivity drift within different humidity ranges in matrix form, enabling systematic modeling and dynamic correction of cross-sensitivity effects. The raw pressure output signal is the electrical signal output by the pressure sensor without any environmental compensation. It is typically expressed in voltage form, such as mV, to represent the pressure value. This signal directly reflects the conversion relationship between the deformation of the pressure sensor's elastic sensitive element and the electrical signal. However, due to environmental factors such as temperature drift and humidity cross-sensitivity, significant measurement errors may occur, requiring correction and compensation through subsequent algorithms. The first pressure correction signal is the pressure signal processed by the dynamic humidity correction algorithm. This signal eliminates the humidity cross-sensitivity effect and provides a preliminary optimized input for temperature compensation.

[0025] The humidity correction module integrates an adaptive humidity compensation algorithm and achieves cross-interference suppression by establishing a cross-sensitivity matrix model between humidity and pressure. This module first divides the humidity range into multiple intervals and uses orthogonal experimental methods to obtain the influence matrix of humidity changes on sensor zero offset and sensitivity drift within each interval. This module then generates a real-time correction coefficient sequence through statistical fitting. During the correction process, a coupled error model of humidity and pressure is constructed to linearly compensate for the systematic deviation caused by humidity in the original output signal of the pressure sensor. At the same time, an inertial filtering mechanism is introduced to eliminate humidity mutation noise, generating a pressure signal that has been preliminarily eliminated from humidity interference, providing an optimized input data source for the temperature compensation process.

[0026] a temperature compensation module, configured to carry an incremental neural compensation network based on transfer learning and build a millisecond-level temperature compensation model based on an embedded edge computing device, and perform temperature compensation on the first pressure correction signal to generate a second pressure compensation signal; It can be understood that the incremental neural compensation network based on transfer learning refers to a neural network model that integrates transfer learning and incremental learning. Transfer learning can leverage the feature extraction capabilities of pre-trained models to transfer general temperature compensation knowledge to specific scenarios, reducing the amount of data required for model training. Incremental learning, on the other hand, uses an online gradient descent algorithm to perform real-time weight updates on newly acquired temperature and pressure data, enabling the model to dynamically adapt to changing operating conditions. This network achieves millisecond-level inference via embedded edge computing devices, performs nonlinear temperature compensation on the humidity-corrected signal, and outputs a second pressure-compensated signal.

[0027] Furthermore, an embedded edge computing device refers to an embedded device used for real-time computing near the data acquisition end, such as an edge node equipped with a GPU or TPU. Its characteristics are low latency and localized decision-making, which enable edge-side deployment and rapid response of the temperature compensation algorithm. The millisecond-level temperature compensation model refers to a real-time compensation model built based on an embedded edge computing device. By optimizing the neural network structure and calculation process, it controls the temperature compensation response time to the millisecond level, which can meet the real-time requirements of dynamic pressure measurement. The second pressure compensation signal refers to the signal processed by the temperature compensation module. This signal further suppresses the temperature drift error and is close to the actual pressure value.

[0028] The temperature compensation module integrates intelligent algorithms and edge computing, and achieves high-precision compensation for temperature drift through neural network models and real-time computing architecture. The module is equipped with an incremental neural compensation network based on transfer learning. It first uses a pre-trained model to extract the common features of temperature and pressure mapping, and then uses an online learning mechanism to update the weights of newly collected data in real time to form a compensation model with dynamic adaptability. The computing architecture of this module adopts embedded edge computing, deploying neural networks at edge nodes close to sensors to achieve real-time compensation calculations with millisecond-level response. At the same time, it aggregates multi-device operation data through the cloud platform and uses federated learning technology to continuously optimize the global model, forming a collaborative architecture of edge real-time computing and cloud model evolution, ensuring the continuous adaptability of the compensation algorithm to complex temperature conditions.

[0029] A fault compensation module is used to automatically trigger the digital twin compensation model to perform fault compensation on the second pressure compensation signal and generate a target pressure output signal when a single sensor node fault in the flexible thermoelectric sensor array is detected.

[0030] The target pressure output signal is the final compensation signal generated by the fault compensation module when a single sensor node in the flexible thermoelectric sensor array fails. This signal provides reliable pressure measurement results for industrial sites. The digital twin compensation model is a virtual simulation model built based on the sensor's physical characteristics and operational data. When a physical sensor node fails, the digital twin model seamlessly replaces the faulty sensor through the synergy of data-driven and physical-driven approaches, achieving compensation and maintaining system measurement accuracy.

[0031] The fault compensation module integrates self-diagnosis mechanisms and virtual simulation technology, enabling seamless switching in the event of sensor failures through a digital twin model. By comparing temperature gradient differences among distributed sensor nodes and combining statistical testing methods, the module can identify abnormal data points and establish a fault detection mechanism. When a single sensor node fails, the digital twin compensation system, consisting of a physical model based on finite element simulation and a data analysis model based on time series prediction, is automatically triggered. The virtual model replaces the faulty sensor for signal deduction, and the compensation signal is generated through collaborative calculations driven by physics and data. This ensures that the system maintains measurement accuracy even in the event of sensor failure, achieving full-link reliability design from hardware to algorithm.

[0032] In the specific implementation process, the thermal expansion coefficient of the elastic sensitive element presents a two-stage negative temperature response performance in the temperature range of -40°C to 125°C, and is less than or equal to 15×10⁻ at 25°C. 6 / ℃.

[0033] The double negative temperature coefficient composite material includes a main material and a negative thermal expansion filler; the main material is a polyetheretherketone-based polymer material, with a corresponding volume proportion of 60% to 70% and a melt index of less than or equal to 15g / 10min; the negative thermal expansion filler is a composite particle of tungsten carbide and carbon fiber, with a corresponding particle size of 5μm to 10μm, a corresponding carbon fiber aspect ratio greater than or equal to 100, and a corresponding volume proportion of 30% to 40%; After the double negative temperature coefficient composite material is compression molded, the temperature coefficient of elastic modulus at 25° C. is less than or equal to -800 MPa / ° C.

[0034] In practical applications, elastic sensitive elements are made of dual negative temperature coefficient composite materials, which exhibit unique thermal expansion characteristics in a wide temperature range. Its thermal expansion coefficient exhibits a two-stage negative temperature response characteristic within the range of -40°C to 125°C, that is, within this temperature range, the material size shrinks as the temperature rises, and exhibits differentiated shrinkage rate characteristics in different temperature ranges. Among them, at a base temperature of 25°C, the thermal expansion coefficient of the element does not exceed 15×10⁻ 6 / ℃, suppressing the interference of temperature changes on the elastic deformation of the pressure sensor from the perspective of material intrinsic properties.

[0035] The double-negative temperature coefficient composite material consists of a main material and a negative thermal expansion filler. The main material is a polyetheretherketone-based polymer material, accounting for 60% to 70% of the volume share. Its melt index index indicates that the material has good thermoplastic processing fluidity. The negative thermal expansion filler is a composite reinforcement particle composed of tungsten carbide and carbon fiber. The tungsten carbide particles are dispersed at the micron scale, and the carbon fiber has an aspect ratio of no less than 100. Together, the two form a three-dimensional network reinforcement structure, accounting for 30% to 40% of the volume.

[0036] The composite material is densified through a compression molding process. After molding, at 25°C, its elastic modulus shows a negative correlation with temperature changes, and the temperature coefficient of the elastic modulus does not exceed -800MPa / °C, that is, the elastic modulus of the material decreases when the temperature rises. This characteristic forms a synergistic compensation effect with the temperature drift law of the pressure sensor, reducing the original temperature drift magnitude of the pressure sensor from a physical mechanism.

[0037] The flexible thermoelectric sensor array includes a polyimide flexible substrate and grid-like distributed nodes; the thickness of the polyimide flexible substrate is 5μm to 15μm, and a micro-electromechanical thermopile sensor is integrated on the surface of the polyimide flexible substrate. The response time of a single micro-electromechanical thermopile sensor is less than or equal to 50ms; the grid-like distributed nodes are integrated with a temperature and voltage conversion unit and a 24-bit analog-to-digital conversion module, supporting temperature field spatial sampling with a resolution of 0.1°C and a sampling frequency greater than or equal to 1000Hz.

[0038] Specifically, the flexible thermoelectric sensor array utilizes a layered integrated architecture, consisting of a flexible substrate and high-density sensing nodes forming a functional complex. The polyimide flexible substrate, which serves as the base, has a micron-level thickness, ensuring mechanical flexibility while providing a high-temperature-resistant, low-deformation support platform for the surface-integrated MEMS thermopile sensors. In the thermopile sensor array fabricated using MEMS technology, individual sensing units exhibit millisecond-level response characteristics, enabling real-time capture of transient changes in the temperature field.

[0039] The grid-like distributed nodes adopt an array topology layout, with each node integrating a temperature-voltage conversion unit and a high-precision analog-to-digital conversion module. The temperature-voltage conversion unit uses the Seebeck effect to convert thermoelectric signals, while the analog-to-digital conversion module uses high-bit quantization to ensure high-precision spatial sampling of the temperature field. This array, through its high-density node distribution, forms a two-dimensional temperature sensing network that supports high-frequency spatial sampling of the temperature field. This allows for precise capture of the temperature gradient distribution characteristics of the measured area, providing high-temporal and spatial resolution ambient temperature data input for subsequent temperature compensation algorithms, meeting the needs of dynamic temperature field monitoring under complex operating conditions.

[0040] The elastic sensitive element and the flexible thermoelectric sensor array are integrated through an integrated packaging process, and the packaging material is an organic silicone rubber with a thermal conductivity greater than or equal to 1.5W / (m·K).

[0041] The elastic sensing element and the flexible thermoelectric sensor array are functionally integrated through an integrated packaging process, using silicone rubber, a material with excellent thermal conductivity, as the encapsulation medium. This integrated packaging process, through precision molding technology, forms a synergistic structural composite of the elastic sensing element and the sensor array, ensuring efficient thermal coupling and mechanical compatibility for temperature field monitoring.

[0042] The silicone rubber used as the encapsulation material has a thermal conductivity of at least 1.5 W / (m·K). This high thermal conductivity ensures that the thermal signal from the temperature field is quickly transmitted to the flexible thermoelectric sensor array, reducing measurement delays and errors caused by thermal resistance. Furthermore, the flexible encapsulation properties of silicone rubber adapt to the deformation requirements of the elastic sensitive element under pressure, avoiding the loss of sensor sensitivity caused by insufficient rigidity or excessive hardening of the encapsulation material.

[0043] Through the coordinated design of the thermal conductivity and mechanical properties of the material, high-precision acquisition of the spatial distribution data of the temperature field is achieved: on the one hand, the high thermal conductivity packaging material ensures the real-time transmission of the temperature signal, enabling the flexible thermoelectric sensor array to accurately capture the temperature gradient changes on the surface of the elastic sensitive element; on the other hand, the integrated packaging process eliminates the air gap thermal resistance and structural gaps in traditional discrete packaging, improving the environmental adaptability and long-term reliability of the system under harsh working conditions such as vibration and shock, and providing structural protection for full-temperature range pressure measurement.

[0044] The humidity correction module is used to integrate a dynamic humidity correction algorithm, establish a cross-sensitivity matrix model of humidity and pressure, correct the original pressure output signal of the pressure sensor, and generate a first pressure correction signal, specifically including: a relationship matrix establishing unit, configured to divide the dynamic humidity data into a plurality of humidity intervals based on the dynamic humidity correction algorithm, and establish a mapping relationship matrix between the humidity value in each humidity interval and the zero drift and sensitivity drift of the original pressure output signal through an orthogonal experimental method; The relationship matrix establishment unit can divide the dynamic humidity data into several humidity intervals based on a dynamic humidity correction algorithm. Then, an orthogonal experimental method can be used to calibrate the sensor output characteristics within each humidity interval, establishing a multidimensional mapping matrix containing humidity values and the zero-point drift and sensitivity drift of the pressure sensor. This method, based on the principle of controlled variables, systematically obtains the quantitative relationship between the sensor zero-point offset and the rate of change of sensitivity under different humidity conditions, forming a basic data matrix that characterizes the cross-sensitivity effect between humidity and pressure, providing parameter support for subsequent dynamic corrections. This matrix, indexed by humidity intervals, stores the interference parameters of humidity changes on the electrical characteristics of the pressure sensor within the corresponding interval, enabling systematic modeling of cross-sensitivity effects.

[0045] a humidity sequence generating unit, configured to dynamically fit the cross-sensitivity matrix model of humidity and pressure based on the mapping relationship matrix using a least squares method and generate a humidity correction coefficient sequence; Among them, the fitting formula of the cross-sensitivity matrix model of humidity and pressure is as follows: Where, Indicates the minimum value operation; Represents the humidity sensitivity correction factor, which is defined as the relative change rate of the pressure sensor's sensitivity when the relative humidity (RH) changes by 1% at a reference temperature of 25°C. Indicates the zero drift correction amount; Indicates the number of humidity calibration samples; Indicates the The measured signal of the pressure sensor under the humidity calibration sample; Indicates the The original pressure output signal of the pressure sensor under the humidity calibration sample; Indicates the Humidity values for each humidity calibration sample.

[0046] Based on the established mapping relationship matrix, the humidity sequence generation unit uses the least squares method to dynamically fit the cross-sensitivity matrix model of humidity and pressure to generate a real-time correction coefficient sequence. Specifically, this unit uses the least squares method to curve fit the pressure sensor output data under the current humidity environment, minimizing the sum of squared errors to determine the zero drift correction and sensitivity correction coefficient under the current humidity conditions.

[0047] This dynamic fitting mechanism adapts to the changing cross-sensitivity characteristics under varying humidity gradients, generating a correction coefficient sequence that matches the real-time humidity environment through iterative calculation. This sequence includes two sets of parameters related to the impact of humidity on the pressure sensor's zero point and sensitivity. Real-time updates ensure the correction model's adaptability to the current environment, resolving the compensation lag problem of traditional fixed-coefficient correction methods under conditions of drastic humidity fluctuations.

[0048] The pressure signal correction unit is used to correct the original pressure output signal based on the humidity correction coefficient sequence to generate the first pressure correction signal, and at the same time introduce a first-order inertial filter to eliminate humidity mutation noise in the first pressure correction signal.

[0049] When the original pressure output signal is corrected based on the humidity correction coefficient sequence to generate the first pressure correction signal, the corresponding calculation formula is as follows: Where, represents a first pressure correction signal; Indicates the original pressure output signal; Indicates humidity value; Indicates temperature value; Indicates the reference sensitivity of the pressure sensor at a reference temperature of 25°C and a relative humidity of 0%. The pressure signal correction unit can use a dynamically generated sequence of humidity correction coefficients to perform linear compensation on the original output signal of the pressure sensor, generating a first pressure correction signal that preliminarily eliminates humidity interference. During the correction process, by constructing a coupled error model of humidity and pressure, the systematic deviation caused by humidity in the original signal is compensated, eliminating the sensor zero point offset and sensitivity fluctuations caused by humidity changes. At the same time, to suppress the high-frequency noise caused by sudden changes in ambient humidity, the unit can introduce a first-order inertial filtering mechanism. By weighted averaging historical data and current data, it smoothes the interference of rapid humidity changes on the correction signal and ensures the stability of the output signal. This composite correction mechanism combines the suppression of humidity cross-sensitivity effects with noise filtering to provide a low-interference, high signal-to-noise ratio pressure signal input for the subsequent temperature compensation module.

[0050] The temperature compensation module is configured to carry an incremental neural compensation network based on transfer learning and to construct a millisecond-level temperature compensation model based on an embedded edge computing device to perform temperature compensation on the first pressure correction signal to generate a second pressure compensation signal, specifically comprising: a temperature matrix generating unit, configured to calculate the spatial gradient of the temperature field based on the temperature spatial distribution data and construct a three-dimensional temperature distribution matrix of the pressure sensor, and introduce the matrix into the incremental neural compensation network based on transfer learning; The calculation formula of the temperature field spatial gradient is as follows: Where, Represents the spatial gradient of the temperature field; Represents a three-dimensional temperature distribution function, which represents the distribution of data points in temperature space and time Temperature value below; and represents the sensor node spacing of the flexible thermoelectric sensor array in the x and y directions; and Represent the gradients of the temperature field in the x and y directions respectively; Represents the distribution of data points in temperature space ( and time Temperature value below; Represents the distribution of data points in temperature space and time Temperature value below; Represents the distribution of data points in temperature space and time Temperature value below; Represents the distribution of data points in temperature space and time The temperature value below.

[0051] It can be understood that the temperature matrix generation unit can realize the digital modeling and feature extraction of the spatial characteristics of the temperature field based on the temperature spatial distribution data. Specifically, the unit performs gradient calculation on the temperature spatial distribution data collected by the flexible thermoelectric sensor array, and uses the numerical differentiation method to solve the rate of change of the temperature field in the two-dimensional plane, and then constructs a three-dimensional data matrix containing the spatiotemporal distribution characteristics of temperature. The matrix uses spatial coordinates and time series as dimensions to store the spatial gradient information and dynamic evolution laws of the temperature field, forming a multi-dimensional feature vector that characterizes the non-uniform distribution characteristics of the temperature field. As the input data source of the neural compensation network, the three-dimensional temperature distribution matrix can convert key parameters such as the spatial gradient characteristics and dynamic change trends of the temperature field into digital features that can be recognized by the algorithm, providing refined ambient temperature representation for subsequent nonlinear temperature compensation.

[0052] A signal temperature compensation unit, configured to construct a millisecond-level temperature compensation model based on an embedded edge computing device, and in combination with the incremental neural compensation network based on transfer learning, perform temperature compensation on the first pressure correction signal to generate a second pressure compensation signal and upload the signal to the cloud server corresponding to the pressure sensor; The signal temperature compensation unit leverages an embedded edge computing architecture to build a real-time compensation model with millisecond-level response capabilities. By deploying a lightweight neural network inference engine within the edge computing device, the incremental neural compensation network based on transfer learning is transformed into a real-time executable computational model. This model uses a first pressure correction signal and a three-dimensional temperature distribution matrix as inputs. Through multi-layer nonlinear mapping, it accurately predicts and compensates for temperature drift errors, generating a second pressure compensation signal that eliminates temperature interference.

[0053] The introduction of an edge computing architecture ensures low-latency compensation calculations, meeting the real-time requirements of dynamic pressure measurement scenarios. Furthermore, the unit can synchronously upload compensated signals to a cloud server, providing distributed data support for global model evolution and forming a collaborative processing model that combines real-time edge computing with cloud data aggregation.

[0054] The temperature compensation model training unit is used to aggregate the second pressure compensation signal through the cloud server using a federated learning algorithm and construct an adversarial sample to train the generalization capability of the millisecond-level temperature compensation model.

[0055] The temperature compensation model training unit can build a closed-loop optimization mechanism for continuous model evolution based on the cloud computing platform. This unit aggregates the compensation data uploaded by distributed edge nodes through the federated learning algorithm, and realizes the collaborative training of multi-source data under the premise of protecting data privacy. During the training process, the model's adaptability to extreme working conditions is enhanced by constructing adversarial samples, and the general compensation knowledge is transferred to specific application scenarios using transfer learning technology. The model is updated in real time based on the newly collected data in combination with the incremental learning mechanism. This hybrid training strategy enables the temperature compensation model to dynamically adapt to complex working conditions such as sudden temperature changes and changes in gradient distribution, and continuously improves the generalization ability and compensation accuracy of the model. The optimized model completed by cloud training is sent to the edge node through differential update to ensure the global performance optimization of the temperature compensation model.

[0056] The fault compensation module is configured to automatically trigger the digital twin compensation model to perform fault compensation on the second pressure compensation signal and generate a target pressure output signal when a single sensor node fault in the flexible thermoelectric sensor array is detected, specifically comprising: a faulty node determining unit, configured to determine a faulty single sensor node by comparing temperature gradient differences between adjacent single sensor nodes in the flexible thermoelectric sensor array and identifying abnormal temperature gradient differences using a Dickson test method; The target signal generation unit is used to automatically trigger the digital twin compensation model to perform fault compensation on the second pressure compensation signal after determining that the single sensor node has a fault, and generate the target pressure output signal, and the digital twin compensation model includes a sensor physical model based on finite element simulation and an operation status prediction model based on LSTM.

[0057] In practical applications, the fault node determination unit can accurately locate the faulty sensor through distributed data collaborative analysis. Specifically, based on the grid topology of the flexible thermoelectric sensor array, the unit can calculate the temperature gradient difference between adjacent sensor nodes in real time and use statistical test methods to identify anomalies in the gradient data sequence. By constructing a difference matrix of the temperature gradients of adjacent nodes, the Dixon test method is used to detect outliers on the data points in the matrix, and the abnormal state of the temperature gradient difference is determined based on the preset significance level, thereby locating the single sensor node with a fault. This fault detection mechanism achieves high-precision identification of faulty nodes through the collaborative perception and statistical analysis of distributed nodes, providing accurate fault location information for subsequent compensation switching.

[0058] After identifying the faulty node, the target signal generation unit triggers the digital twin compensation mechanism, achieving seamless restoration of the measurement signal. The digital twin compensation model consists of two sub-models: a physics-driven and a data-driven model. These include a sensor physics model based on finite element simulation and an LSTM-based operating status prediction model. The finite element simulation-based sensor physics model predicts the sensor's theoretical output by inputting real-time temperature field distribution data through a virtual simulation system coupled with thermal-structural coupling. The LSTM (Long Short-Term Memory)-based operating status prediction model learns from the temporal characteristics of historical operating data to infer sensor output trends under current operating conditions. These two models generate a compensation signal using a weighted fusion strategy to correct for measurement deviations in the second pressure compensation signal caused by the sensor failure, ultimately outputting a target pressure output signal with full spatial resolution. This composite compensation mechanism combines the advantages of sensor physics simulation and data-driven prediction, ensuring that the system maintains measurement accuracy even in the event of a single sensor failure, enabling a seamless transition from hardware failure to algorithmic compensation.

[0059] Example 2 like Figure 2 As shown, an intelligent pressure sensor temperature compensation method, the method comprising: S1, reduces the original temperature drift magnitude of the pressure sensor through the elastic sensitive element; S2, collecting temperature spatial distribution data and dynamic humidity data of the pressure sensor through a flexible thermoelectric sensor array; S3, establishing a cross-sensitivity matrix model of humidity and pressure based on a dynamic humidity correction algorithm, and correcting the original pressure output signal of the pressure sensor to generate a first pressure correction signal; S4, using an incremental neural compensation network based on transfer learning and building a millisecond-level temperature compensation model based on an embedded edge computing device, to perform temperature compensation on the first pressure correction signal to generate a second pressure compensation signal; S5. When a single sensor node failure in the flexible thermoelectric sensor array is detected, the digital twin compensation model is automatically triggered to perform fault compensation on the second pressure compensation signal to generate a target pressure output signal.

[0060] Through the introduction of the above embodiments, the present invention uses an intelligent pressure sensor temperature compensation system to reduce the original temperature drift magnitude of the pressure sensor through elastic sensitive elements; collects temperature spatial distribution data and dynamic humidity data of the pressure sensor through a flexible thermoelectric sensor array; based on the dynamic humidity correction algorithm, establishes a cross-sensitive matrix model of humidity and pressure, corrects the original pressure output signal of the pressure sensor, and generates a first pressure correction signal; through an incremental neural compensation network based on transfer learning, a millisecond-level temperature compensation model is constructed based on an embedded edge computing device, and the first pressure correction signal is temperature compensated to generate a second pressure compensation signal; when a single sensor node failure is detected in the flexible thermoelectric sensor array, the digital twin compensation model is automatically triggered to compensate for the fault of the second pressure compensation signal and generate a target pressure output signal, thereby achieving precise control of pressure measurement errors in the entire temperature range and improving the adaptability and intelligence level of the system in harsh environments.

[0061] The present invention can physically suppress temperature drift through the elastic sensitive element of the double negative temperature coefficient composite material, reduce the sensitivity of the sensor to ambient temperature changes, lay the hardware foundation for high-precision measurement, and capture the spatial gradient of the temperature field through a flexible thermoelectric sensor array, breaking through the limitations of traditional single-point temperature measurement, enabling the system to accurately perceive the temperature distribution characteristics in complex environments, and providing comprehensive environmental data support for the compensation algorithm. The system of the present invention forms an intelligent processing mechanism of first correction and then compensation through a dynamic humidity correction algorithm and an incremental neural compensation network based on transfer learning, effectively eliminating the cross-sensitivity effect of humidity and temperature. At the same time, it achieves real-time response through embedded edge computing and relies on the cloud platform to achieve continuous model evolution, so that the system has the ability to dynamically adapt to different working conditions. By integrating the digital twin compensation mode, the present invention can seamlessly switch when a single sensor fails, ensuring the stable operation of the system under abnormal conditions. The system of the present invention constructs a multi-level compensation system to achieve precise control of pressure measurement errors in the entire temperature range, significantly improving the adaptability and reliability of the sensor in harsh environments such as aerospace and industrial automation, and providing an efficient solution for precision measurement scenarios.

[0062] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0063] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, including a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0064] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

Claims

1. An intelligent pressure sensor temperature compensation system, characterized in that: The system comprises: A hardware sensing module, including an elastic sensitive element made of a double negative temperature coefficient composite material, is used to reduce the original temperature drift magnitude of the pressure sensor; An environmental data acquisition module, comprising a flexible thermoelectric sensor array, arranged on the surface of the pressure sensor, for acquiring temperature spatial distribution data and dynamic humidity data of the pressure sensor; a humidity correction module, configured to integrate a dynamic humidity correction algorithm, establish a cross-sensitivity matrix model of humidity and pressure, correct the original pressure output signal of the pressure sensor, and generate a first pressure correction signal; a temperature compensation module, configured to carry an incremental neural compensation network based on transfer learning and build a millisecond-level temperature compensation model based on an embedded edge computing device, and perform temperature compensation on the first pressure correction signal to generate a second pressure compensation signal; A fault compensation module is used to automatically trigger the digital twin compensation model to perform fault compensation on the second pressure compensation signal and generate a target pressure output signal when a single sensor node fault in the flexible thermoelectric sensor array is detected.

2. The intelligent pressure sensor temperature compensation system according to claim 1, characterized in that: The thermal expansion coefficient of the elastic sensitive element presents a two-stage negative temperature response performance in the temperature range of -40°C to 125°C, and is less than or equal to 15×10⁻ at 25°C. 6 / ℃.

3. The intelligent pressure sensor temperature compensation system according to claim 1, characterized in that: The double negative temperature coefficient composite material includes a main material and a negative thermal expansion filler; the main material is a polyetheretherketone-based polymer material, with a corresponding volume proportion of 60% to 70% and a melt index of less than or equal to 15g / 10min; the negative thermal expansion filler is a composite particle of tungsten carbide and carbon fiber, with a corresponding particle size of 5μm to 10μm, a corresponding carbon fiber aspect ratio greater than or equal to 100, and a corresponding volume proportion of 30% to 40%; After the double negative temperature coefficient composite material is compression molded, the temperature coefficient of elastic modulus at 25° C. is less than or equal to -800 MPa / ° C.

4. The intelligent pressure sensor temperature compensation system according to claim 1, characterized in that: The flexible thermoelectric sensor array includes a polyimide flexible substrate and grid-like distributed nodes; the thickness of the polyimide flexible substrate is 5μm to 15μm, and a micro-electromechanical thermopile sensor is integrated on the surface of the polyimide flexible substrate. The response time of a single micro-electromechanical thermopile sensor is less than or equal to 50ms; the grid-like distributed nodes are integrated with a temperature and voltage conversion unit and a 24-bit analog-to-digital conversion module, supporting temperature field spatial sampling with a resolution of 0.1°C and a sampling frequency greater than or equal to 1000Hz.

5. The intelligent pressure sensor temperature compensation system according to claim 1, characterized in that: The elastic sensitive element and the flexible thermoelectric sensor array are integrated through an integrated packaging process, and the packaging material is an organic silicone rubber with a thermal conductivity greater than or equal to 1.5W / (m·K).

6. The intelligent pressure sensor temperature compensation system according to claim 1, characterized in that: The humidity correction module is used to integrate a dynamic humidity correction algorithm, establish a cross-sensitivity matrix model of humidity and pressure, correct the original pressure output signal of the pressure sensor, and generate a first pressure correction signal, specifically including: a relationship matrix establishing unit, configured to divide the dynamic humidity data into a plurality of humidity intervals based on the dynamic humidity correction algorithm, and establish a mapping relationship matrix between the humidity value in each humidity interval and the zero point drift and sensitivity drift of the original pressure output signal through an orthogonal experimental method; a humidity sequence generating unit, configured to dynamically fit the cross-sensitivity matrix model of humidity and pressure based on the mapping relationship matrix using a least squares method and generate a humidity correction coefficient sequence; The pressure signal correction unit is used to correct the original pressure output signal based on the humidity correction coefficient sequence to generate the first pressure correction signal, and at the same time introduce a first-order inertial filter to eliminate humidity mutation noise in the first pressure correction signal.

7. The intelligent pressure sensor temperature compensation system according to claim 1, characterized in that: The temperature compensation module is configured to carry an incremental neural compensation network based on transfer learning and to construct a millisecond-level temperature compensation model based on an embedded edge computing device to perform temperature compensation on the first pressure correction signal to generate a second pressure compensation signal, specifically comprising: a temperature matrix generating unit, configured to calculate the spatial gradient of the temperature field based on the temperature spatial distribution data and construct a three-dimensional temperature distribution matrix of the pressure sensor, and introduce the three-dimensional temperature distribution matrix into the incremental neural compensation network based on transfer learning; a signal temperature compensation unit, configured to construct the millisecond-level temperature compensation model based on the embedded edge computing device, and in combination with the incremental neural compensation network based on transfer learning, perform temperature compensation on the first pressure correction signal to generate the second pressure compensation signal and upload the signal to the cloud server corresponding to the pressure sensor; The temperature compensation model training unit is used to aggregate the second pressure compensation signal through the cloud server using a federated learning algorithm and construct an adversarial sample to train the generalization capability of the millisecond-level temperature compensation model.

8. The intelligent pressure sensor temperature compensation system according to claim 1, characterized in that: The fault compensation module is configured to automatically trigger the digital twin compensation model to perform fault compensation on the second pressure compensation signal and generate a target pressure output signal when a single sensor node fault in the flexible thermoelectric sensor array is detected, specifically comprising: a faulty node determining unit, configured to determine a faulty single sensor node by comparing temperature gradient differences between adjacent single sensor nodes in the flexible thermoelectric sensor array and identifying abnormal temperature gradient differences using a Dickson test method; The target signal generation unit is used to automatically trigger the digital twin compensation model to perform fault compensation on the second pressure compensation signal after determining that the single sensor node has a fault, and generate the target pressure output signal, and the digital twin compensation model includes a sensor physical model based on finite element simulation and an operation status prediction model based on LSTM.

9. An intelligent pressure sensor temperature compensation method, applied to an intelligent pressure sensor temperature compensation system according to any one of claims 1 to 8, the method comprising: Reduce the original temperature drift magnitude of the pressure sensor by using elastic sensitive elements; Collecting temperature spatial distribution data and dynamic humidity data of the pressure sensor through a flexible thermoelectric sensor array; Based on a dynamic humidity correction algorithm, a cross-sensitivity matrix model of humidity and pressure is established to correct the original pressure output signal of the pressure sensor to generate a first pressure correction signal; By using an incremental neural compensation network based on transfer learning and building a millisecond-level temperature compensation model based on an embedded edge computing device, the first pressure correction signal is temperature compensated to generate a second pressure compensation signal; When a single sensor node failure in the flexible thermoelectric sensor array is detected, the digital twin compensation model is automatically triggered to perform fault compensation on the second pressure compensation signal to generate a target pressure output signal.

Images