Temperature and pressure double-variable wide-temperature-zone transmitting system

Abstract

The invention relates to the technical field of sensing and measurement and control, and discloses a temperature and pressure double-variable wide-temperature-zone transmitting system, which comprises a composite sensing acquisition module used for synchronously acquiring analog signals of temperature and pressure in a unified sensitive structure, and a signal processing module used for carrying out analog-to-digital conversion, normalization and derivative calculation on the analog signals so as to obtain a temperature and pressure double-variable wide-temperature-zone transmitting signal. The system comprises a neural network modeling module used for acquiring temperature and pressure variables and constructing feature vectors, a coupling modeling module used for acquiring a coupling matrix and an offset vector between the temperature and pressure variables through neural network modeling, and a variable decoupling module used for carrying out inverse transformation on the feature vectors according to the coupling matrix and the offset vector. According to the invention, the temperature sensing unit and the pressure sensing unit are integrated in the same sensitive structure, a composite sensing diaphragm is constructed, and bivariate synchronous acquisition with consistent height in space is realized. According to the technology, the consistency of multiple physical quantity measuring points is improved. Compared with a traditional separated measurement framework, the problems of large displacement error and asynchronous response among multiple sensors are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sensing and measurement and control, in particular to a temperature and pressure dual-variable wide-temperature-range transmission system. BACKGROUND

[0002] In the application fields of industrial process control, power system state monitoring and intelligent sensing, temperature and pressure as two kinds of basic physical quantities often need to be monitored and transmitted simultaneously. The existing industrial transmission system usually completes the measurement and signal conversion of each variable through a temperature transmitter and a pressure transmitter respectively, and then integrates them into an upper data processing platform. This kind of dual-variable sensing system is widely used in pipe network operation, electrical equipment thermal state judgment and energy efficiency diagnosis, etc. It puts forward higher requirements on sensing range, accuracy and reliability, especially in the case of severe environmental temperature fluctuation or limited operation space, which puts forward new challenges to the integration degree and adaptability of the system.

[0003] However, in the prior art, there are still many limitations and technical bottlenecks, such as the traditional temperature and pressure sensing adopts a discrete structure arrangement, which has poor spatial consistency, and the physical offset exists when the sensing node is deployed, resulting in the incoordination of sensing data in space and time. At the same time, most systems establish a coupling model between variables based on linear compensation or lookup table method, which is difficult to effectively represent the nonlinear response characteristics in high temperature or extreme environment. In addition, the fixed sampling frequency processing method collects all state changes at the same frequency without distinction, which causes waste of energy consumption and data bandwidth, and the energy utilization efficiency of edge devices is not high. In the data communication part, the periodic data reporting mechanism is generally used, which is difficult to balance the quick response of key events and the low load operation of the system, especially in the Internet of Things environment, there is a lack of flexibility and optimization ability between network resource utilization rate and communication strategy.

[0004] Therefore, the present application provides a temperature and pressure dual-variable wide-temperature-range transmission system to solve the problems of the prior art. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides a temperature and pressure dual-variable wide-temperature-range transmission system to solve the problems of difficult synchronous collection of dual variables in complex temperature range, insufficient coupling and decoupling precision, low system energy efficiency and lack of dynamic scheduling of data transmission.

[0006] To achieve the above purpose, the present application realizes the following technical scheme: a temperature and pressure dual-variable wide-temperature-range transmission system, comprising:

[0007] A composite sensing and collecting module is used to synchronously acquire analog signals of temperature and pressure in a unified sensitive structure;

[0008] A signal processing module is used to perform analog-to-digital conversion, normalization and derivative calculation on the analog signals, and construct a feature vector;

[0009] a coupling modeling module configured to obtain a coupling matrix and a bias vector between temperature and pressure variables through neural network modeling;

[0010] a variable decoupling module configured to perform inverse transformation on a feature vector according to the coupling matrix and the bias vector, and output decoupled temperature estimation and pressure estimation;

[0011] a scheduling and communication module configured to adjust a sampling frequency according to a variable change trend, and upload the estimations and system states to an external device, the scheduling and communication module further comprising a prediction module configured to predict a feature vector at a next time and generate a scheduling factor accordingly.

[0012] Preferably, the composite perception acquisition module comprises:

[0013] a ceramic-metal composite film sensitive structure suitable for conducting environmental physical fields in a wide temperature range;

[0014] a piezoresistive pressure perception unit and a platinum resistance temperature perception unit, respectively outputting analog voltage signals corresponding to physical quantities.

[0015] Preferably, the feature vector constructed by the signal processing module is:

[0016]

[0017] wherein, represents a normalized value of a pressure signal at the kth sampling point; represents a normalized value of a temperature signal at the kth sampling point; ΔV p (k) represents a first derivative of the normalized value of the pressure signal, calculated as: ΔV t (k) represents a first derivative of the normalized value of the temperature signal; Δt represents a sampling period time interval.

[0018] Preferably, the coupling modeling module outputs a coupling matrix M k and a bias vector b k , satisfying the following mapping relationship:

[0019] y k =M k ·z k +b k ;

[0020] wherein, y k is a normalized observation vector obtained by the signal processing module, containing and z k is a hidden state variable vector, respectively a normalized pressure estimation and a normalized temperature estimation Mk is a two-dimensional coupling matrix obtained by modeling; b k is a bias vector.

[0021] Preferably, the variable decoupling module is configured to decouple variables, satisfying the inverse transformation relationship:

[0022]

[0023] wherein, represents the inverse matrix of the coupling matrix M k ; y k is a current normalized observation vector; b k is a bias vector; z k is a decoupled variable estimation result, including and

[0024] Preferably, the scheduling and communication module includes a scheduling control unit, which controls the data sampling rhythm based on a scheduling factor f k , wherein the sampling period Δt k satisfies:

[0025]

[0026] wherein, Δt k represents the current sampling time interval; f k represents the current calculated scheduling frequency factor.

[0027] Preferably, the scheduling frequency factor is generated by a prediction module in the scheduling and communication module, satisfying the following calculation relationship:

[0028]

[0029] wherein, f max represents the maximum sampling frequency; λ is a response factor; is a predicted next time feature vector; x k is a current feature vector; ||·||2 is the Euclidean norm.

[0030] Preferably, the scheduling and communication module is configured to upload the estimated value, the current sampling frequency and the system state, and receive the model parameters or update instructions sent by the remote device.

[0031] Preferably, the scheduling and communication module includes a fault detection unit, which is configured to trigger a redundant reporting, a parameter locking or a reset operation when a system drift, an estimation anomaly or a communication interruption occurs.

[0032] The present application provides a temperature and pressure dual-variable wide-temperature-range transmission method, comprising the following steps:

[0033] Collecting temperature and pressure analog signals;

[0034] Analog-to-digital conversion, normalization and first derivative calculation are performed on the analog signals to construct a feature vector;

[0035] The feature vector is input into a neural network model to obtain a coupling matrix and a bias vector;

[0036] The estimated values of temperature and pressure are obtained by inverse transformation decoupling;

[0037] Based on the predicted feature changes, a scheduling factor is generated, and the sampling frequency is adjusted;

[0038] The estimated values and system state are uploaded to an external device.

[0039] The present application provides a temperature and pressure dual-variable wide-temperature-zone transmission system, which has the following advantages:

[0040] 1. The present application integrates temperature and pressure sensing units in the same sensitive structure, constructs a composite sensing diaphragm, and realizes highly consistent dual-variable synchronous acquisition in space. This technology improves the consistency of multi-physical quantity measurement points. Compared with traditional separate measurement architecture, it solves the problems of large displacement error and asynchronous response among multiple sensors.

[0041] 2. The present application uses a neural network dynamic output temperature and pressure coupling matrix, combined with online feature calculation, to construct an integrated mechanism of coupling modeling and decoupling estimation. This scheme enhances the adaptive ability of the model to nonlinear coupling. Traditional linear fitting methods have a significant decrease in accuracy under wide temperature zone application, and the present application avoids the temperature zone failure problem of such methods.

[0042] 3. The scheduling mechanism based on change rate prediction is introduced in the present application, which no longer relies on fixed frequency sampling, but dynamically regulates through derivative change trend. The result is a significant improvement in system energy efficiency. Compared with the existing continuous equal period sampling method, it significantly reduces power consumption and data redundancy, and improves the endurance performance of edge devices.

[0043] 4. The present application uses event-driven data reporting through the communication module, combined with multi-interface protocol output, which greatly improves the real-time performance and network adaptability of data transmission. Unlike traditional fixed period upload structure, it solves the problem of key data delay and unnecessary bandwidth occupation restriction, and is especially suitable for low-power Internet of Things scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 The system architecture diagram of the present application;

[0045] Figure 2 The method flowchart of the present application. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the specification of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0047] Please refer to the drawings in the specification of the present application Figure 1 The embodiment of the present application provides a temperature and pressure dual-variable wide-temperature-range transmission system, comprising:

[0048] A composite sensing and collecting module is configured to synchronously acquire analog signals of temperature and pressure in a unified sensitive structure.

[0049] In the embodiment, the composite sensing and collecting module is configured to realize synchronous sensing of pressure and temperature variables at the same physical space position, thereby serving as a data source module of the system of the present application and providing basic data support for subsequent signal processing, coupling modeling, variable decoupling and remote communication modules.

[0050] To realize synchronous acquisition of temperature and pressure variables, the composite sensing and collecting module preferably adopts an integrated packaging structure, and different types of sensing elements are spatially coupled and integrated through a composite sensitive diaphragm, so that the two variables are simultaneously sensed at the same measuring point. The structure not only ensures the consistency of the variable sampling points, but also provides a spatial consistency physical basis for subsequent modeling and decoupling processing.

[0051] In the preferred embodiment of the present application, the composite sensing structure comprises a ceramic-metal composite diaphragm, and the diaphragm forms a sensitive layer with high mechanical strength and uniform thermal conductivity through a hot-pressing sintering process, and is used to jointly bear temperature sensing and pressure sensing units.

[0052] The pressure sensing part preferably adopts a piezoresistive sensor structure, which realizes response to the pressure acting on the surface of the diaphragm through a stress sensing unit. The unit forms a strain bridge structure inside, which can convert a small deformation response into a voltage signal output, and the output signal is denoted as V p .

[0053] The temperature sensing part can adopt a platinum resistance type temperature sensing element, which realizes sensing of a temperature signal based on the positive correlation between resistance and temperature. The temperature element is connected through a bridge mode, and in combination with a driving circuit, an analog signal related to the change of the ambient temperature is output, and the output signal is denoted as V t .

[0054] To ensure that the sensing module has stable response capability in a wide temperature range (e.g. low to high temperature range), the metal layer in the ceramic-metal composite film material preferably has good thermal expansion matching performance to reduce the mechanical mismatch problem caused by different thermal expansion coefficients. This structure helps to improve the stability and long-term working life of the sensing device.

[0055] In terms of signal path design, the composite sensing acquisition module transmits V p and V t to the subsequent signal processing module at the same time. The lead-out structure can include a filter network with temperature compensation function for preliminary balancing of interference suppression and drift control of the analog signal before sampling.

[0056] In terms of workflow, when the external environment applies pressure and temperature changes, the pressure and temperature signals on the sensing diaphragm are received by the two sensing units at the same time, which are converted into voltage signals V p and V t , and output to the signal processing module in synchronous timing. Since the sensing units are integrated in the same composite film structure, the two physical quantities have spatial consistency, which is beneficial to the subsequent construction of a multivariate modeling system based on coupling relationship.

[0057] It is worth noting that the design of this module avoids the mutual interference of temperature and pressure sensing paths in structure, and effectively reduces the influence of additional stress caused by thermal gradient on pressure signal by reasonable arrangement of sensing unit and signal lead position, thereby improving the physical independence of signal and the subsequent decoupling accuracy.

[0058] The composite sensing acquisition module in this embodiment can also adapt to the needs of different sampling periods, and work with the scheduling factor of the signal processing module to realize dynamic adaptive adjustment of data sampling frequency. In the system initialization stage, the acquisition module triggers the sampling process through the control end, and synchronously outputs the analog signals of pressure and temperature within the set period.

[0059] In engineering implementation, the composite structure facilitates batch manufacturing through encapsulation modular design, supports high consistency, stable performance and miniaturized integration requirements, and is suitable for practical needs of embedded multivariate sensing system.

[0060] In summary, the composite sensing acquisition module provides key underlying data support for multivariate coupled sensing and model-driven decoupling calculation of the system by integrating synchronous sensing of temperature and pressure signals, co-film structure design and stability design of signal transmission path, and lays a technical foundation for subsequent processing flow.

[0061] The signal processing module is used for analog-to-digital conversion, normalization and derivative calculation of the analog signal, and construction of a feature vector.

[0062] In this embodiment, the signal processing module is used for digitizing the raw analog signals output by the composite perception acquisition module, constructing feature expressions, and dynamically generating scheduling factors, which constitutes a key intermediate link in the system.

[0063] The signal processing module receives two analog signals from the composite perception acquisition module, which are pressure perception V p and temperature perception signal V t To facilitate subsequent modeling calculations and system integration, the analog signals need to be first converted by an analog-to-digital conversion operation to convert continuous signals into discrete time series. The analog-to-digital conversion can be achieved through an optimal ADC chip, which is connected to the control logic and periodically sampled based on the system clock.

[0064] After completing the analog-to-digital conversion, the resulting discrete signals still have dimensional differences and non-uniform numerical scales. To overcome the uneven effects of different signal amplitudes on modeling, the signal processing module performs a normalization operation on the sampled data. Preferably, a linear normalization method based on the maximum and minimum value range is used, and the specific expression is as follows:

[0065]

[0066] wherein, represents the normalized kth sampling value; V i,min and V i,max represent the minimum and maximum values of the corresponding signals in the initialization or sliding window; V i (k) is the original analog voltage signal sampled at the kth time; i∈{p,t} is the index, indicating whether the current signal is processing the pressure (p) or temperature (t) channel. To further characterize the variable change trend and dynamic characteristics, the signal processing module introduces a first-order differential derivative calculation mechanism based on the normalization result. By taking the difference between the normalized values of the previous and subsequent sampling times and dividing by the sampling period Δt, an approximate change rate index can be obtained, and the specific calculation formula is as follows:

[0067]

[0068] This derivative quantity represents the instantaneous change rate of the variable, which has a direct effect on judging the variable fluctuation intensity and scheduling frequency.

[0069] After completing the above preprocessing operations, the signal processing module constructs a multi-dimensional feature vector as the input for subsequent modeling and scheduling. The feature vector x k includes the normalized pressure value, normalized temperature value, and first-order derivative information at the current time, and the specific structure is as follows:

[0070]

[0071] wherein, represents the normalized value of the pressure signal at the kth sampling point; represents the normalized value of the temperature signal at the kth sampling point; ΔV p (k) represents the first derivative of the normalized value of the pressure signal, calculated as: ΔV t (k) represents the first derivative of the normalized value of the temperature signal; Δt represents the sampling period time interval.

[0072] The feature vector not only retains the amplitude information of the original variable, but also introduces a derivative component reflecting the dynamic change characteristics, facilitating the subsequent module to learn and reason the variable coupling characteristics.

[0073] In the present embodiment, the signal processing module is further provided with a scheduling factor generation unit for dynamically adjusting the frequency control parameters in the data acquisition and communication process. The introduction of the scheduling factor is to cope with the different variable change states and avoid unnecessary high-frequency sampling or communication of system resources in the variable stable stage.

[0074] The scheduling mechanism is based on the change trend of the feature vector sequence to realize prediction, preferably using a sliding window mechanism to construct a historical sequence x k = {x k-L+1 ,…,x k}, and using an embedded prediction model to generate a predicted value of the feature vector at the next time After obtaining the difference between the current predicted value and the actual value, the Euclidean distance is calculated as a quantitative indicator of the variable change amplitude, defined as follows:

[0075]

[0076] wherein, represents the predicted change quantity (or feature error vector) at the kth time, representing the difference between the prediction result of the next time state by the system and the current actual state; is the predicted feature vector for the k+1th time, obtained by the prediction module (such as a small neural network or a predictor) based on the current information; x k is the current feature vector collected at the kth time, obtained by the sensor sampling and processing (such as feature combination after normalization and derivative calculation).

[0077]

[0078] wherein, d k represents the Euclidean norm of the feature change quantity at the kth time; is the predicted change vector at the kth time, obtained by differencing the predicted feature vector and the current feature vector; is the two-norm, that is, the square root of the sum of the squares of the components of the vector; is the two-norm, that is, the square root of the sum of the squares of the components of the vector; is the summation symbol from the 1st dimension to the 4th dimension, because the feature vector x is 4-dimensional (usually containing: pressure normalized value, temperature normalized value, pressure derivative, temperature derivative); is the predicted value of the jth feature component at the k+1th moment; x j (k) is the actual sampling value of the jth feature component at the kth moment.

[0079] Finally, the scheduling factor f k can be calculated by the following formula:

[0080] f k = f max ·(1-exp(-λ·d k ));

[0081] wherein, λ is an adjustable response factor; f max is the maximum settable frequency; exp(·) represents the exponential function. The above formula realizes the dynamic adjustment logic that the sampling frequency increases when the variable changes sharply, and the sampling frequency decreases when the variable is stable.

[0082] The output of the signal processing module in the embodiment includes: the constructed feature vector x k , as the input of the coupling modeling module and the variable decoupling module; and the scheduling factor f k , as the frequency control basis of the communication module and the data reporting logic.

[0083] The signal processing module is deployed in the intermediate processing unit of the system of the application, and can be realized by using an embedded microprocessor or a digital signal controller. The core algorithm module is loaded by a system firmware calling mode, and is high-frequency coordinated with the data acquisition and model inference process.

[0084] The signal processing module described in the embodiment realizes effective conversion between physical variables and mathematical modeling through multiple sub-processes such as normalization processing, first derivative calculation, feature vector construction and scheduling factor generation, and provides input basis for dynamic regulation and control of system resources, and is an important component in the multivariable sensing and intelligent transmission system of the application.

[0085] The coupling modeling module is used to obtain the coupling matrix and bias vector between the temperature and pressure variables through neural network modeling;

[0086] In this embodiment, the coupling modeling module is used to construct the nonlinear coupling relationship between the temperature variable and the pressure variable, and the coupling relationship is expressed in a parameterized structure for subsequent variable decoupling and error compensation processing. The module takes the feature vector output by the signal processing module as input, and dynamically generates the coupling mapping function under the current state through the embedded and executable lightweight modeling structure.

[0087] This embodiment adopts a neural network structure as a coupling modeling tool, which outputs a parameter set for constructing a mapping relationship by inputting the feature vector at the current sampling time. This network structure is suitable for embedded chip environment, has low computational complexity and limited storage requirements, and is convenient for integration and application in actual industrial control devices.

[0088] The input feature vector is denoted as:

[0089]

[0090] wherein, and are the normalized voltage values of the current time pressure and temperature respectively; ΔV p (k) and ΔV t (k) are the first derivatives thereof, used to represent the instantaneous change trend. The neural network preferably adopts a fully connected feedforward structure, including an input layer, at least one hidden layer, and an output layer. The input layer receives the above-mentioned 4-dimensional feature vector, the hidden layer performs nonlinear transformation on the input through an activation function, and the output layer gives an explicit representation of the coupling model parameters. This structure is not limited to a specific network depth or activation function selection, but in implementation, it is preferred to use an activation method such as ReLU which has lower computational cost in embedded inference. The output of the network includes a set of numerical values for constructing matrix M k and vector b k , a total of six scalar parameters. This parameter set constitutes the coupling mapping model at the current time, and its structure is as follows:

[0091]

[0092] wherein, is the coupling transformation matrix, describing the cross-influence relationship between the two variables under the current system state; m ij (k) represents the element of the i-th row and j-th column of the matrix, which is usually a time step k-dependent coefficient, used to describe how to linearly transform the state or variable; is the bias term, used to correct the system deviation caused by linear mapping; b1(k), b2(k) are the components of the bias vector in the 1st and 2nd dimensions respectively. It is used to introduce external inputs or control quantities of the system.

[0093] The parameters are dynamically generated by the neural network at runtime, have certain input state sensitivity, and can reflect the mapping rule in the variable change mode in real time, thereby improving the adaptability of the system in the non-steady state process.

[0094] To ensure the stability and convergence of the model output, the neural network can be trained offline by historical data before system deployment, and the training target is to minimize the mean square error between the model prediction output and the known calibration sample. The trained network weight can be stored in the device storage unit, and the output is calculated in real time by the fixed inference process during actual operation.

[0095] In the actual operation process, when the system receives the feature vector x k generated by the signal processing module, it immediately inputs it into the neural network for forward inference to output the coupling matrix M k and the bias vector b k The model output does not require complex iteration and can be completed by one round of forward propagation, which is suitable for resource-constrained terminal environments.

[0096] After the coupling relationship is constructed, the related parameters are synchronously transmitted to the variable decoupling module for subsequent decoupling mapping and physical variable estimation operations. This structure changes the modeling process from "explicit function construction" to "parameter generation driven", which has stronger flexibility and portability.

[0097] It should be noted that the coupling modeling module in this embodiment does not directly output the predicted value of the physical variable, but outputs the coupling coefficient set used to calculate the physical variable. This can reduce the dependence of the modeler output on other modules of the system, and also facilitates modular decoupling and function division in system design.

[0098] In the above manner, the coupling modeling module realizes modeling and expression of the coupling relationship between multiple variables, and its output structure provides a parameter basis for subsequent decoupling and estimation, which is a key link in the coupling-decoupling processing mechanism in the present application.

[0099] The variable decoupling module is used to perform inverse transformation on the feature vector according to the coupling matrix and the bias vector, and output the decoupled temperature estimate and pressure estimate;

[0100] In this embodiment, the variable decoupling module is used to perform parameterized mapping inverse transformation based on the constructed multi-variable coupling model, to realize independent restoration of physical quantities in the coupling scenario. The module receives the coupling parameter set output by the coupling modeling module and the observed feature vector provided by the signal processing module, and combines matrix inversion and vector operation, etc. linear algebra methods to decouple and reconstruct the temperature and pressure variables.

[0101] In the scheme, the structure of the coupling modeling module output is a mapping matrix M k and a bias vector b k As mentioned previously:

[0102]

[0103] The structure describes the coupling projection relationship between the pressure signal and the temperature signal in the current system state. The variable decoupling module needs to inversely deduce the uncoupled expression of the original physical quantity from the coupling model to ensure that the true value of each physical parameter can be accurately reconstructed when there is interference or superposition between different variables. To achieve the above goal, the variable decoupling module first receives the original normalized perception voltage vector, denoted as:

[0104]

[0105] wherein, represents the normalized perception voltage in the main direction (d-axis); is the normalized perception voltage in the vertical direction (q-axis).

[0106] The above observation vector can be regarded as the current coupling output of the system, and the corresponding modeling input is a set of implicit physical state variables z k , denoted as:

[0107]

[0108] wherein, and are the dimensionless estimates of the reconstructed temperature and pressure variables, respectively.

[0109] In this embodiment, the coupling mapping process is modeled as the following linear transformation:

[0110] y k = M k ·z k +b k ;

[0111] The variable decoupling module accordingly establishes the corresponding inverse transformation to solve the unknown variable z k in the above equation. This process inversely deduces the estimation result of the original variable by performing matrix inversion on the coupling matrix M k and algebraic operation combined with the bias vector:

[0112]

[0113] wherein, represents the matrix M kinverse of the matrix M, provided that the matrix is invertible at the time of calculation. To improve system stability, the variable decoupling module is provided with a limiting condition judgment mechanism for detecting the invertibility of the matrix M k . Preferably, when the absolute value of the determinant |M k | is less than a certain set threshold, a backup processing path is triggered or the decoupling parameters of the last state are maintained to prevent the introduction of system instability factors when the matrix is singular. The decoupling estimation result Although it is a normalized dimensionless variable, it still maintains a one-to-one mapping relationship with the actual physical quantity. Therefore, the variable decoupling module can further introduce a normalized inverse transformation to restore the corresponding physical quantity unit expression:

[0114]

[0115] where T min ,T max ,P min ,P max are the upper and lower boundary values of the temperature and pressure variables estimated in the historical samples. The inverse normalization process can complete parameter loading in the system initialization phase, or can be dynamically updated according to the sliding window to enhance the system adaptive ability.

[0116] The core of the variable decoupling module is to accurately extract the values of individual physical variables from the coupled mapping relationship. Its complete process includes input vector reception, matrix algebra calculation, boundary condition judgment, and numerical domain mapping, etc. Each processing process is periodically triggered and executed in the system running period, and is synchronized with the data reporting module.

[0117] In addition, considering the difference of hardware resources, the variable decoupling module in this embodiment can adopt multiple configuration methods in the form of implementation. For example, for resource-constrained edge devices, fixed-point operations can be used instead of floating-point matrix inversion, and lookup table method can be used to realize normalized inverse transformation; for system platforms with hardware floating point support, the complete linear algebra processing path can be retained to realize higher precision variable estimation.

[0118] Through the above-mentioned manner, the variable decoupling module realizes effective disassembly of the coupled information in the composite signal and restores the estimated values of the respective true physical variables, providing high-reliability technical support for multi-variable perception, error compensation and data reporting of the system, and is an important link for realizing the coupling modeling and intelligent perception closed loop in the present application.

[0119] The scheduling and communication module is used for adjusting the sampling frequency according to the variable change trend, and uploading the estimated value and system state to an external device. The scheduling and communication module further comprises a prediction module for predicting the feature vector at the next time and generating a scheduling factor accordingly.

[0120] In this embodiment, the scheduling control and data reporting module is used to control the system acquisition rhythm, data caching strategy and reporting mechanism based on the dynamic scheduling factor generated by the signal processing module and the physical quantity estimation results obtained by the variable decoupling module, so as to achieve a dynamic balance between system resource consumption and perception accuracy, and support system-level communication and remote data synchronization.

[0121] The scheduling control part firstly calculates the scheduling factor f from the signal processing module. k Establish a collection trigger frequency control mechanism. The scheduling factor, generated in the previous step based on the feature variation, has a certain degree of timing sensitivity and responsiveness to changing trends. This scheduling factor is considered an indicator of the current system's sampling update rate. The control module combines this factor with the internal clock to generate the collection trigger signal.

[0122] To ensure the accuracy and programmability of scheduling control, a soft timer mechanism is introduced in this embodiment, whose timing period is dynamically set by the following function:

[0123]

[0124] Where, Δt k is the current sampling period; f k is the scheduling factor output by the signal processing module at the previous moment. The control logic sets the soft timer to count up Δt k When the time is up, the acquisition trigger signal is generated to trigger the composite perception module to complete the next round of data sampling and restart the timing cycle to realize the real-time driving effect of the scheduling factor.

[0125] During the operation of the scheduling mechanism, the module needs to maintain dynamic cache management of the current state to avoid cache overflow or loss caused by adjusting the scheduling frequency too quickly or too slowly. To this end, it is preferred to organize memory resources in a ring buffer manner, and set up an expiration and elimination mechanism to release low-value data to ensure memory efficiency.

[0126] The data reporting part receives the temperature and pressure estimation values ​​(T k ,P k ), and combined with context information such as system status identifier, timestamp, and scheduling factor, a complete data message structure is constructed. This structure includes but is not limited to the following fields:

[0127] Timestamp identification field, recording the time when data was generated;

[0128] Variable valuation field, containing the decoupled physical variable T k ,P k ;

[0129] Scheduling information field, containing the current scheduling factor f k , sampling period Δt k ;

[0130] State byte field, used to mark the current sampling quality, system load, etc. running information.

[0131] The data packet adopts a preferred field alignment method in the structure design, so as to facilitate protocol analysis and compatible processing between different platforms.

[0132] In the data reporting process, in order to reduce the communication bandwidth consumption and avoid redundant information transmission, the system introduces an event-driven data sending mechanism. It determines whether the trigger condition is met according to the variable change trend, and only when the physical variable change amplitude exceeds a certain set threshold or the system scheduling frequency changes significantly, the one-time data reporting process is started. The event determination logic is realized based on the following criteria:

[0133] |T k -T k-1 |>∈ T or |P k -P k-1 |>∈ P ;

[0134] Where, ∈ T , ∈ P are the threshold factors of temperature and pressure change, which can be set by initialization or updated adaptively during running.

[0135] For the time that does not meet the event condition, the data reporting module still maintains the background cache mechanism, writes the corresponding variable state into the local buffer area, to support subsequent data audit, abnormal backtracking or batch synchronization.

[0136] At the communication protocol level, the data reporting module supports multiple interface methods, including but not limited to serial communication, CAN bus, wireless link, etc. The specific interface implementation can be selected according to the application scene, and the low-power communication methods such as UART or SPI are preferred in embedded platforms, and the transmission mechanisms based on Modbus, MQTT or custom protocol stack are used in industrial networking scenarios.

[0137] In this embodiment, the scheduling control and data reporting module is at the end execution layer in the system structure, directly connected with the perception hardware and the host computer or server end, and undertakes the task of final data delivery. At the same time, the multi-dimensional control of scheduling rhythm, cache strategy and communication mechanism also ensures the dynamic balance and reliability guarantee of resource use in the system running process.

[0138] Overall, the scheduling control and data reporting module realizes intelligent rhythm adjustment based on variable state, efficient data transmission based on events and unified integration of multi-source state information, and is a key end unit for realizing the multi-variable intelligent sensing closed loop in the application.

[0139] Please refer to the accompanying Figure 2 The application also provides a temperature and pressure dual-variable wide-temperature-range transmission method, including the following steps:

[0140] Collecting temperature and pressure analog signals;

[0141] Performing analog-digital conversion, normalization and first-order derivative calculation on the analog signals to construct a feature vector;

[0142] Inputting the feature vector into a neural network model to obtain a coupling matrix and a bias vector;

[0143] Obtaining temperature and pressure estimated values by inverse transformation decoupling;

[0144] Generating a scheduling factor based on predicted feature changes and adjusting a sampling frequency;

[0145] Uploading the estimated values and system state to an external device.

[0146] The system in the embodiment can be used to execute the method embodiments, and has similar principles and technical effects, which will not be described here.

[0147] Although the embodiments of the application have been shown and described, it is to be understood that the application is not limited to these embodiments. It will be obvious to those skilled in the art that various changes, modifications, substitutions and alterations can be made thereto without departing from the spirit and scope of the application, which are defined by the appended claims and their equivalents.

Claims

1. Temperature and pressure dual variable wide temperature range transmission system, characterized by: include: Composite sensing acquisition module, used to synchronously acquire analog signals of temperature and pressure in a unified sensitive structure; A signal processing module, configured to perform analog-to-digital conversion, normalize and calculate derivatives of the analog signal, and construct a feature vector; A coupling modeling module is used to obtain the coupling matrix and bias vector between temperature and pressure variables through neural network modeling; a variable decoupling module, configured to perform an inverse transformation on the eigenvector according to the coupling matrix and the bias vector, and output a decoupled temperature estimate and a pressure estimate; The scheduling and communication module is used to adjust the sampling frequency according to the trend of variable changes and upload the estimated value and system status to the external device. The scheduling and communication module also includes a prediction module for predicting the feature vector at the next moment and generating a scheduling factor based on it.

2. The temperature and pressure dual variable wide temperature range transmission system according to claim 1 is characterized in that: The composite sensing acquisition module includes: Ceramic-metal composite film sensitive structure, suitable for conducting environmental physical fields in a wide temperature range; The piezoresistive pressure sensing unit and the platinum resistance temperature sensing unit respectively output analog voltage signals corresponding to physical quantities.

3. The temperature and pressure dual variable wide temperature range transmission system according to claim 1, characterized in that: The characteristic vector constructed by the signal processing module is: in, represents the normalized value of the pressure signal at the kth sampling point; Indicates the normalized value of the temperature signal at the kth sampling point; ΔV p (k) represents the first-order derivative of the normalized value of the pressure signal, which is calculated as: ΔV t (k) represents the first derivative of the normalized value of the temperature signal; Δt represents the sampling period time interval.

4. The temperature and pressure dual variable wide temperature range transmission system according to claim 1, characterized in that: The coupling modeling module output includes the coupling matrix M k With the bias vector b k , satisfying the following mapping relationship: y k =M k ·z k +b k ; Among them, y k is the normalized observation vector obtained by the signal processing module, including and z k is the hidden state variable vector, which is the normalized pressure estimate and temperature estimate respectively M k is the two-dimensional coupling matrix obtained by modeling; b k is the bias vector.

5. The temperature and pressure dual variable wide temperature range transmission system according to claim 1, characterized in that: The variable decoupling module is used to decouple variables to satisfy the following inverse transformation relationship: in, Denotes the coupling matrix M k The inverse matrix of y k is the current normalized observation vector; b k is the bias vector; z k is the variable estimation result obtained by decoupling, including and 6. The temperature and pressure dual variable wide temperature range transmission system according to claim 1, characterized in that: The scheduling and communication module includes a scheduling control unit, which is based on the scheduling factor f k Control data sampling rhythm, the sampling period Δt k satisfy: Where Δt k Indicates the current sampling time interval; f k Indicates the currently calculated scheduling frequency factor.

7. The temperature and pressure dual variable wide temperature range transmission system according to claim 6, characterized in that: The scheduling frequency factor is generated by the prediction module in the scheduling and communication module and satisfies the following calculation relationship: Among them, f max represents the maximum sampling frequency; λ is the response factor; is the predicted feature vector for the next moment; x k is the current eigenvector; ||·||2 is the Euclidean norm.

8. The temperature and pressure dual variable wide temperature range transmission system according to claim 1, characterized in that: The scheduling and communication module is used to upload estimated values, current sampling frequency and system status, and receive model parameters or update instructions sent by remote devices.

9. The temperature and pressure dual variable wide temperature range transmission system according to claim 8, characterized in that: The scheduling and communication module includes a fault detection unit for triggering redundancy reporting, parameter locking or resetting operations when system drift, estimation anomaly or communication interruption occurs.

10. A method for transmitting dual variables of temperature and pressure over a wide temperature range, applied to a system for transmitting dual variables of temperature and pressure over a wide temperature range as claimed in any one of claims 1 to 9, characterized in that: The following steps are involved: Collect temperature and pressure analog signals; Performing analog-to-digital conversion, normalization, and first-order derivative calculation on the analog signal to construct a feature vector; Input the feature vector into the neural network model to obtain the coupling matrix and bias vector; The estimated values ​​of temperature and pressure are obtained by decoupling through inverse transformation; Generate scheduling factors based on predicted feature changes and adjust sampling frequency; Upload estimates and system status to external devices.

Images