A method, device, storage medium and measuring system for calibrating a sensor system environment compensation

Abstract

The application discloses a kind of free calibration sensing system environment compensation methods, comprising: constructing measurement matrix and environment matrix;Real value vector and influence coefficient matrix are solved by least residual method;Compensation after calculating measured value matrix;Reliability inspection;And when inspection passes, the real value vector of influence coefficient matrix calculated by having been obtained is followed.This application another aspect discloses a kind of free calibration sensing system environment compensation measurement system, including measurement sensing device, calculation verification unit and environmental variable monitoring device, measurement sensing device and environmental variable monitoring device are in a preset time window and are measured multiple times synchronously, and calculation verification unit includes compensation module and verification evaluation module.The sensing system environment compensation method of the application does not need to be calibrated in advance in fixed environment, and environmental influence coefficient can be dynamically estimated by real-time measurement data and real-time dynamic compensation is carried out.

Description

A calibration-free environmental compensation method, device, storage medium, and measurement system for a sensing system. Technical Field

[0001] This invention relates to the field of sensing and measurement systems, and particularly to a calibration-free method, apparatus, storage medium, and measurement system for environmental compensation of sensing systems. Background Technology

[0002] High-precision measurement systems, such as those composed of sensors measuring displacement, pressure, temperature, flow rate, and concentration, are typically calibrated under specific standard environmental parameters (e.g., room temperature of 25°C). When environmental variables (temperature, humidity, air pressure, etc.) change dynamically (e.g., temperature fluctuations due to diurnal temperature variations), the measured values ​​will exhibit systematic deviations due to environmental influences. Existing environmental compensation methods usually require pre-calibration under different environmental conditions before establishing an environmental compensation model, resulting in high equipment commissioning costs and response lag. Therefore, how to avoid complex pre-calibration and dynamically perform environmental compensation under the influence of multiple environmental variables is a problem that urgently needs to be solved. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to perform multiple synchronous measurements of various quantities and environmental variables within a preset time window, decompose the measurement results into a part that remains constant within the time window (true value), a part that changes linearly with environmental variables, and a noise part, calculate relevant parameters using the minimum residual method, and perform real-time dynamic compensation, thereby avoiding complex pre-calibration work.

[0004] To achieve the above objectives, the present invention provides a calibration-free environmental compensation method for a sensing system, comprising the following steps:

[0005] S1: Obtain m synchronous measurement data. Each synchronous measurement data includes n measurement quantities and p environmental variables. Construct an n×m dimensional measurement matrix X with the measurement quantities and a p×m dimensional environmental matrix E with the environmental variables. m, n and p are positive integers, and m>1.

[0006] S2: Solve for the n-dimensional true value vector μ and the n×p-dimensional influence coefficient matrix β based on the measurement matrix X and the environment matrix E;

[0007] S3: Calculate the compensated measurement matrix using the true value vector μ and the influence coefficient matrix β.

[0008] S4: Perform a reliability check. If the check fails, trigger a parameter update. If the check passes, proceed to the next step S5.

[0009] S5: Use the acquired influence coefficient matrix β for measurement compensation: acquire m synchronous measurement data, each synchronous measurement data includes n measurement quantities and p environmental variables, construct an n×m dimensional measurement matrix X with the measurement quantities, construct a p×m dimensional environmental matrix E with the environmental variables, calculate the true value vector μ based on the measurement matrix X, the environmental matrix E and the acquired influence coefficient matrix β, and return to step S3 for looping.

[0010] Preferably, step S2 specifically includes:

[0011] Solve by combining formulas (5) and (7) to obtain the n-dimensional true value vector μ and the n×p-dimensional influence coefficient matrix β:

[0012]

[0013] In the formula: i = 1…n, j = 1…m, k = 1…p; the dimensions of the true value vector μ are μ i μ i This represents the true value corresponding to the i-th measurement; the dimensions of the influence coefficient matrix β are respectively β ik ,β ik This represents the influence coefficient of the k-th environmental variable on the i-th measurement; the dimensions of the measurement matrix X are xi, xj ... ij x ij This represents the value of the i-th measurement during the j-th synchronous measurement; each dimension of the environment matrix E is e kj e kj This represents the value of the k-th environmental variable during the j-th synchronous measurement; 1 m It is an m-dimensional column vector of all 1s;

[0014] In step S5, the true value vector μ is calculated based on the measurement matrix X, the environmental matrix E, and the obtained influence coefficient matrix β. Specifically, the true value vector μ is calculated according to formula (5).

[0015] Preferably, step S3 specifically involves: calculating the compensated measurement value matrix according to formula (8).

[0016]

[0017] In the formula E new This refers to the environment matrix corresponding to the current state or the environment matrix corresponding to a preset reference state.

[0018] Preferably, in step S4, the reliability test includes environmental change constraint test and residual statistical constraint test.

[0019] Preferably, the environmental change constraint test includes testing whether the environmental matrix E satisfies formula (9). If the environmental matrix E satisfies formula (9), the test passes.

[0020]

[0021] In the formula Δe k This is the preset threshold for the change of the kth environmental variable.

[0022] Preferably, the environmental change constraint test includes statistically analyzing the cumulative change values ​​of each environmental variable over a continuous preset time period. If the cumulative change value does not exceed the corresponding preset change threshold, the test is passed.

[0023] Preferably, the residual statistical constraint test includes at least one of the T-test, chi-square test, or Grubbs test.

[0024] Preferably, the residual statistical constraint test includes the following steps:

[0025] Define the nm-dimensional residual vector r according to formula (10):

[0026]

[0027] In the formula, i = 1…n, j = 1…m;

[0028] For s = 1…nm, r s If all conditions are met, the test is passed.

[0029]

[0030] In the formula: α is the preset significance level; t α / 2 (df) represents the two-tailed quantile of the T-distribution; df represents the degrees of freedom; df = nm - np. For residual standard deviation estimation,

[0031] Preferably, the trigger parameter update includes: optimizing the influence coefficient matrix β through gradient descent, genetic algorithm or online learning, and / or returning to step S1.

[0032] Another aspect of the present invention provides a device for environmental compensation of a calibration-free sensing system, comprising a data integration module, a compensation module, a reliability verification module, and a parameter update module; the device is an electronic device with data processing capabilities, and the device is used to implement the aforementioned sensing system environmental compensation method;

[0033] The data integration module is used to receive raw measurement data and real-time environmental monitoring data, perform time-series correlation between the raw measurement data and real-time environmental monitoring data through a preset data alignment algorithm and convert them into synchronous measurement data, and construct a measurement matrix X and an environmental matrix E based on the synchronous measurement data;

[0034] The compensation module is used to calculate the true value vector μ, the influence coefficient matrix β, and the compensated measurement value matrix.

[0035] The reliability verification module is used to perform reliability verification; the parameter update module is used to optimize and update the influence coefficient matrix β when the parameter update is triggered.

[0036] Preferably, the device is an embedded processor or an application-specific integrated circuit, wherein the embedded processor includes an ARM processor or an FPGA processor.

[0037] In another aspect, the present invention provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the aforementioned sensing system environmental compensation method.

[0038] Another aspect of the present invention provides a measurement system for environmental compensation of a calibration-free sensing system, comprising a measurement sensing device, a calculation and verification unit, and an environmental variable monitoring device;

[0039] The measurement sensing device includes at least one sensor for acquiring raw measurement data; the measurement sensing device and the calculation verification unit are data connected.

[0040] The environmental variable monitoring device includes at least one sensor for acquiring real-time environmental monitoring data; the environmental variable monitoring device and the calculation and verification unit are data-connected.

[0041] Both the measurement and sensing device and the environmental variable monitoring device complete m synchronous measurements within a preset time window;

[0042] The computational verification unit includes a computational compensation module and a verification and evaluation module;

[0043] The calculation and compensation module is used to calculate intermediate parameters without calibration based on the data obtained from m synchronous measurements, and to compensate for the measured values ​​based on the intermediate parameters.

[0044] The verification and evaluation module is used to verify and evaluate the intermediate parameters, and triggers the update of the intermediate parameters when the verification and evaluation fails.

[0045] Preferably, the computational verification unit is an embedded processor or an application-specific integrated circuit.

[0046] Preferably, the computational verification unit is implemented by an executable program in an edge computing device or a cloud server cluster.

[0047] Preferably, the verification and evaluation module uses at least one of the statistical error index, confidence index, distribution characteristic index, and dynamic performance index to verify and evaluate the intermediate parameters.

[0048] Preferably, the calculation verification unit is used to implement the aforementioned sensing system environmental compensation method.

[0049] In summary, the present invention has the following beneficial effects:

[0050] (1) The sensing system environmental compensation method of the present invention does not need to be pre-calibrated in a fixed environment. It can dynamically estimate the environmental impact coefficient through real-time measurement data and perform real-time dynamic compensation.

[0051] (2) The sensing system environmental compensation method of the present invention supports modeling the linear influence of multiple environmental variables on multiple measurements, and the measurement system is suitable for complex environmental scenarios.

[0052] (3) The reliability test of the sensing system environmental compensation method of the present invention can be constrained by the residual statistical constraint test and the environmental change constraint test, which can trigger parameter updates in a timely manner and ensure the reliability of compensation. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figures 1 and 2 are flowcharts of the environmental compensation method for the sensing system according to an embodiment of the present invention;

[0055] Figure 3 is a structural diagram of the measurement system according to an embodiment of the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] The sensing system environmental compensation method disclosed in this application supports linear influence modeling of multiple environmental variables and multiple measurements. The measured quantities are the raw measurement data of the target quantity obtained through the measurement sensing unit, including but not limited to displacement, pressure, temperature, flow rate, and concentration. The environmental variables are real-time environmental monitoring data obtained through the environmental variable monitoring unit. Common environmental variables include, but are not limited to:

[0058] (1) Meteorological parameters: temperature, humidity, air pressure, light intensity, wind speed, wind direction, rainfall, snowfall, ultraviolet radiation intensity, and concentration of air pollutants (such as PM2.5, PM10, CO2, NO). x SO2), radiation levels (such as gamma rays and ultraviolet radiation), and lightning activity frequency;

[0059] (2) Geographic and topographic parameters: geographical location (latitude and longitude, altitude), topographic slope, soil moisture, soil composition, groundwater level, water flow velocity (such as the flow velocity of fluids in rivers and pipes), and tidal data;

[0060] (3) Physical environment parameters: electromagnetic field strength, vibration frequency, noise level (decibels), mechanical stress, thermal radiation, sound wave propagation characteristics, pressure gradient;

[0061] (4) Chemical and biological parameters: water quality parameters (pH value, conductivity, dissolved oxygen, etc.), chemical gas concentration (such as H2S, NH3, VOCs), microbial concentration, biological pollution indicators;

[0062] (5) Special environmental parameters: seismic activity data, geological subsidence rate, corrosive gas concentration, radiation source intensity, particulate matter size distribution, and material aging degree;

[0063] The measurement system performs m synchronous measurements within a preset short time window Δt. Each synchronous measurement acquires initial measurement data and real-time environmental monitoring data, including raw data corresponding to n measured quantities and p environmental variables. The sensing system environmental compensation method disclosed in this application mathematically models this data based on the following three fundamental assumptions:

[0064] (i) Assumption of constant true value: The same measurement quantity consists of the true value and the measurement error, wherein the true value remains unchanged in multiple synchronous measurements within a certain time window;

[0065] (II) Error decomposition assumption: The measurement error consists of independent Gaussian noise and linear environmental influences. The influence of each environmental variable on the measured quantity has a linear response relationship and can be linearly superimposed to form the total environmental influence.

[0066] (iii) Assumption of gradual environmental change: The changes in each environmental variable are within the corresponding preset threshold range in each two adjacent measurements, and the nonlinear effect can be ignored.

[0067] The following table lists the definitions of some of the mathematical symbols used in the modeling:

[0068] Table 1 Definitions of Mathematical Symbols

[0069]

[0070] Where 1≤i≤n, 1≤j≤m, 1≤k≤p.

[0071] Based on the aforementioned basic assumptions, the value of the i-th measurement quantity during the j-th synchronous measurement can be decomposed into three parts according to formula (1):

[0072]

[0073] Where μ i The true value to be estimated; For the linear influence term of environmental variables; n ij It is Gaussian noise, and its mean is 0 after m measurements.

[0074] Formula (1) can be written in matrix form as formula (2):

[0075]

[0076] Among them 1 m Given an m-dimensional column vector of all ones, transpose it to obtain... Given an m-dimensional all-1 row vector, multiplying the true value vector by this all-1 row vector on the left yields a constant value matrix of the true values.

[0077] Construct the residual sum of squares objective function R(μ,β) according to formula (3):

[0078]

[0079] This function characterizes the overall deviation between the measured value and the model prediction value, and the optimization objective of the least squares method is to solve for minR(μ,β).

[0080] The objective function R(μ,β) of the residual sum of squares is applied to μ. i Find the partial derivatives of (i = 1…n) and set them to zero, i.e., formula (4):

[0081]

[0082] The formula is obtained by rearranging:

[0083]

[0084] The objective function R(μ,β) of the sum of squared residuals is applied to β. ik (i=1…n,k=1…p) Taking the partial derivatives and setting them to zero, we get formula (6):

[0085]

[0086] Using matrix differentiation, the optimal solution can be obtained as shown in formula (7):

[0087] β=(XE T -μ1 m E T (EE) T ) -1 (7);

[0088] Since both formulas (5) and (7) are linear relationships containing the true value vector μ and the influence coefficient matrix β, they can be combined to obtain a system of linear equations, which can be solved by the matrix method to obtain the corresponding true value vector μ and influence coefficient matrix β.

[0089] Since this embodiment does not require calibration, and therefore no standard state is set, the current state can be used as a reference state during compensation. Historical measurement data is dynamically compensated to the current environmental state. During linear dynamic compensation, the compensated measurement value matrix is ​​calculated according to formula (8).

[0090]

[0091] In the formula E new Substitute the environmental matrix corresponding to the current environmental state into the compensated measurement matrix. This is equivalent to the measured value after eliminating system noise under the current environmental conditions.

[0092] In other embodiments, those skilled in the art may also use an environment matrix E corresponding to a preset reference state. 0 As E new Compensation can be performed using a benchmark; alternatively, a reference state can be used as the relative origin of environmental variables, i.e., a linear mapping can be applied to environmental variables to make E 0 =0, at which point the true value is the corresponding compensated measurement value.

[0093] In a preferred embodiment, those skilled in the art can also further refine the compensated measurement matrix using dynamic compensation methods such as nonlinear mapping or adaptive filtering, depending on the application scenario.

[0094] To ensure the robustness of mathematical modeling, the model can be reliably tested and adaptively updated based on the test results: if the test passes, subsequent measurements and compensation can use the obtained influence coefficient matrix β and calculate the true value vector μ according to formula (5), simplifying the calculation process; if the test fails, parameter updates are triggered, and the influence coefficient matrix β can be optimized through gradient descent, genetic algorithms, or online learning, or formulas (5) and (7) can be re-established in subsequent measurements and compensation for solution. Reliability testing includes testing of environmental change constraints and residual statistical constraints. In a preferred embodiment, other models can also be introduced, and the output data of each model can be compared through multi-model cross-validation, triggering parameter updates or model switching based on the comparison results.

[0095] (I) Environmental Change Constraints

[0096] To ensure the linearity assumption of the model holds, the assumption of gradual environmental change must be satisfied, i.e., formula (9) must be satisfied:

[0097]

[0098] In the formula Δe k This is a preset threshold for the change of the kth environmental variable, which can be determined through hardware filtering or historical data statistics.

[0099] In other embodiments, environmental change constraints can also be tested using a sliding window statistical method, that is, the cumulative change value of each environmental variable is counted within a continuous preset time period (e.g., 24 hours), and the test passes if the cumulative change value does not exceed the change threshold.

[0100] In a preferred embodiment, those skilled in the art can also convert the raw data obtained from the environmental variable monitoring unit into mapped environmental variables that better conform to the linear assumption through a preset mapping relationship (e.g., nonlinear mapping).

[0101] (II) Residual Statistical Constraints

[0102] The variance, skewness, p-value, and other statistics can be tested using methods such as the T-test, chi-square test, and Grubbs test. Taking the T-test as an example, the residual vector r is defined according to formula (10):

[0103]

[0104] In the formula, i = 1…n, j = 1…m, so the residual vector r is nm dimensional.

[0105] For s = 1…nm, r s If all conditions are met, the test is passed.

[0106]

[0107] In the formula: α is the preset significance level; t α / 2 (df) represents the two-tailed quantile of the T-distribution; df represents the degrees of freedom; df = nm - np. For residual standard deviation estimation,

[0108] In summary, the flow chart of the environmental compensation method according to the embodiments of this application is shown in Figures 1 and 2, and includes the following steps:

[0109] S1: Acquire m synchronous measurement data. Each synchronous measurement data includes n measurements and p environmental variables for which raw data is acquired simultaneously. The acquisition of raw data is completed within a preset time window Δt. It can be assumed that the measurements do not change within the time window Δt. Construct an n×m dimensional measurement matrix X with the measurements and a p×m dimensional environmental matrix E with the environmental variables.

[0110] S2: Solve the true value vector μ and the influence coefficient matrix β by combining formulas (5) and (7);

[0111] S3: Calculate the compensated measurement matrix using the true value vector μ and the influence coefficient matrix β.

[0112] S4: Perform a reliability check. If the check passes, proceed to step S5. If the check fails, trigger a parameter update. When updating the parameters, you can jump to step S6 for optimization (Figure 1) and / or return to step S1 to remodel and obtain parameters (Figure 2).

[0113] S5: Use the acquired influence coefficient matrix β for measurement compensation: acquire m synchronous measurement data, each synchronous measurement data includes n measurement quantities and p environmental variables, construct an n×m dimension measurement matrix X with the measurement quantities, and construct a p×m dimension environmental matrix E with the environmental variables; calculate the true value vector μ according to formula (5); return to step S3 to loop;

[0114] S6: Optimize the influence coefficient matrix β through gradient descent, genetic algorithm or online learning, and return to the loop of steps S3 to S5 after achieving the optimization goal.

[0115] If other models are introduced to perform compensation calculations simultaneously, multi-model cross-validation is performed during reliability testing, and the most suitable compensation result is selected for output through multi-model cross-validation.

[0116] Through the above steps, this embodiment achieves calibration-free environmental compensation, which means that the intermediate parameters required for the compensation method can be directly calculated from the current environmental conditions or several synchronous measurement data in historical measurements, and further environmental compensation can be performed without calibrating the method under standard environmental conditions.

[0117] Figure 3 illustrates the measurement system structure for environmental compensation of a sensing system according to an embodiment of this application, including a measurement sensing device 2, a calculation and verification unit 3, and an environmental variable monitoring device 4. The measurement system is installed as an integrated device on the gas storage tank 1. The measurement sensing device 2 is a displacement gauge used to detect displacement changes at the measurement point A. The measurement sensing device 2 and the calculation and verification unit 3 are connected, and the measurement sensing device 2 transmits raw measurement data to the calculation and verification unit 3. The environmental variable monitoring device 4 includes multiple sensors that detect environmental data such as temperature, humidity, air pressure, and light intensity. The environmental variable monitoring device 4 is connected to the calculation and verification unit 3, and the environmental variable monitoring device 4 transmits real-time environmental monitoring data to the calculation and verification unit 3. Both the measurement sensing device 2 and the environmental variable monitoring device 4 complete m synchronous measurements within a preset short time window Δt.

[0118] In other embodiments, depending on the needs of the actual application scenario, the sensors that can be set in the environmental variable monitoring device 4 include, but are not limited to: temperature sensors (such as thermistors and thermocouples), humidity sensors (such as capacitive and resistive types), atmospheric pressure sensors (such as piezoresistive and capacitive types), light intensity sensors (such as photodiodes and photoresistors), electromagnetic field sensors (such as Hall effect sensors), vibration sensors (such as accelerometers and piezoelectric sensors), noise sensors (such as microphone arrays), geolocation modules (such as GPS and BeiDou), water quality sensors (such as pH electrodes and conductivity probes), and gas sensors (such as electrochemical sensors and photoionization detectors).

[0119] In a preferred embodiment, the calculation verification unit 3 may be an embedded processor (such as ARM, FPGA, etc.) or an application-specific integrated circuit (ASIC) or other hardware data processing device disposed on the measurement sensing device 2, used to implement the aforementioned sensing system environmental compensation method; the calculation verification unit 3 includes a data integration module, a compensation module, a reliability verification module, and a parameter update module; the data integration module is used to receive the raw measurement data and real-time environmental monitoring data, perform time-series correlation between the raw measurement data and real-time environmental monitoring data through a data alignment algorithm and convert them into synchronous measurement data, and construct a measurement matrix X and an environmental matrix E based on the synchronous measurement data; the data alignment method includes, but is not limited to, timestamp synchronization, frequency matching, or time-series correlation based on a physical model; the compensation module is used to calculate the true value vector μ, the influence coefficient matrix β, and the compensated measurement value matrix. The reliability verification module is used to perform reliability verification; the parameter update module is used to optimize and update the influence coefficient matrix β when parameter updates are triggered.

[0120] In a preferred embodiment, the verification of environmental change constraints can be achieved through a hardware sensor threshold alarm. That is, when an environmental variable changes significantly, the environmental variable monitoring device 4 provides a corresponding threshold alarm signal to the calculation verification unit 3, and the reliability verification module triggers parameter updates based on the threshold alarm signal.

[0121] In a preferred embodiment, the functionality of the computational verification unit 3 is implemented by an executable program in a computer device such as an edge computing device or a cloud server cluster, replacing a dedicated hardware data processing device. The computational verification unit 3 includes a computational compensation module and a verification evaluation module. The measurement sensing device 2 and the environmental variable monitoring device 4 are respectively connected to the edge computing device or cloud server cluster to form the entire measurement system. The computational compensation module is used to calculate intermediate parameters (e.g., the true value vector μ and influence coefficient matrix β in the aforementioned embodiment, or other mathematical models, machine learning, etc., can be used to calculate intermediate parameters) without calibration based on the data obtained from synchronous measurement, and to compensate for the measured values ​​based on the intermediate parameters. The verification evaluation module can perform verification evaluation using one of the following indicators, triggering parameter updates when verification fails, thereby achieving dynamic adaptive updates of the intermediate parameters, including but not limited to:

[0122] (1) Statistical error indicators: variance, standard deviation, mean square error (MSE), maximum absolute error (MAE);

[0123] (2) Confidence indicators: confidence interval, confidence level, statistical significance test (such as t test, chi-square test, Grubbs test);

[0124] (3) Distribution characteristic indicators: skewness, kurtosis, probability density function (PDF) or cumulative distribution function (CDF) of error distribution;

[0125] (4) Dynamic performance indicators: temporal stability of data before and after compensation (such as autocorrelation coefficient), correlation coefficient matrix between environmental variables and error.

[0126] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A calibration-free environmental compensation method for a sensing system, characterized in that, Includes the following steps: S1: Obtain m synchronous measurement data points. Each synchronous measurement data point includes n measurements and p environmental variables. Construct a system based on the measurements. A measurement matrix X of dimension is constructed using the environmental variables. An environment matrix E is given, where m, n, and p are positive integers, and m >

1. S2: Solve for the n-dimensional true value vector μ based on the measurement matrix X and the environment matrix E. The influence coefficient matrix β of dimension; S3: Calculate the compensated measurement matrix using the true value vector μ and the influence coefficient matrix β. S4: Perform a reliability check. If the check fails, trigger a parameter update; if the check passes, proceed to step S5. S5: Perform measurement compensation using the acquired influence coefficient matrix β: acquire m synchronous measurement data points, each containing n measurements and p environmental variables, and construct a... A measurement matrix X of dimension is constructed using the environmental variables. The environmental matrix E is used to calculate the true value vector μ based on the measurement matrix X, the environmental matrix E, and the obtained influence coefficient matrix β. Then, the process returns to step S3 and repeats.

2. The sensing system environmental compensation method according to claim 1, characterized in that, Step S2 specifically involves solving the n-dimensional true value vector μ by simultaneously applying formulas (5) and (7). The influence coefficient matrix β of dimension: （5）； (7); where: i = 1 … n, j = 1 … m, k = 1 … p; the dimensions of the true value vector μ are respectively ， This represents the true value corresponding to the i-th measurement; the dimensions of the influence coefficient matrix β are respectively... ， This represents the influence coefficient of the k-th environmental variable on the i-th measurement; the dimensions of the measurement matrix X are respectively... ， This represents the value of the i-th measurement during the j-th synchronous measurement; the dimensions of the environment matrix E are respectively... ， This represents the value of the k-th environmental variable during the j-th synchronous measurement; It is an m-dimensional column vector of all 1s; in step S5, the true value vector μ is calculated based on the measurement matrix X, the environment matrix E and the obtained influence coefficient matrix β. Specifically, the true value vector μ is calculated according to formula (5).

3. The environmental compensation method for a sensing system according to claim 1, characterized in that, Step S3 specifically involves calculating the compensated measurement matrix according to formula (8). ： (8); In the formula This refers to the environment matrix corresponding to the current state or the environment matrix corresponding to a preset reference state.

4. The sensing system environmental compensation method according to claim 1, characterized in that, In step S4, the reliability test includes environmental change constraint test and residual statistical constraint test.

5. The environmental compensation method for a sensing system according to claim 4, characterized in that, The environmental change constraint test includes testing whether the environmental matrix E satisfies formula (9). If the environmental matrix E satisfies formula (9), the test passes. (9); where: Δe k The threshold value for the change of the k-th environmental variable is preset; the dimensions of the environmental matrix E are respectively... ， This represents the value of the k-th environmental variable during the j-th synchronous measurement.

6. The environmental compensation method for a sensing system according to claim 4, characterized in that, The environmental change constraint test includes statistically analyzing the cumulative change values ​​of various environmental variables over a continuous preset time period. If the cumulative change value does not exceed the corresponding preset change threshold, the test is passed.

7. The environmental compensation method for a sensing system according to claim 4, characterized in that, The residual statistical constraint test includes at least one of the following: T-test, chi-square test, or Grubbs test.

8. The environmental compensation method for a sensing system according to claim 4, characterized in that, The residual statistical constraint test includes the following steps: Define an nm-dimensional residual vector r according to formula (10): (10); where: i = 1 … n, j = 1 … m; the dimensions of the true value vector μ are respectively ， This represents the true value corresponding to the i-th measurement; the dimensions of the influence coefficient matrix β are respectively... ， This represents the influence coefficient of the k-th environmental variable on the i-th measurement; the dimensions of the measurement matrix X are respectively... ， This represents the value of the i-th measurement during the j-th synchronous measurement; the dimensions of the environment matrix E are respectively... ， Let r represent the value of the k-th environmental variable at the j-th synchronous measurement; for s = 1 … nm, r s If all conditions are met, then the test is passed: (11); where α is the preset significance level; Here, denoted as the two-sided quantile of the T-distribution; df represents the degrees of freedom. ； For residual standard deviation estimation, 。 9. The environmental compensation method for a sensing system according to claim 1, characterized in that, The trigger parameter update includes: optimizing the influence coefficient matrix β through gradient descent, genetic algorithm or online learning, and / or returning to step S1.

10. A device for environmental compensation of a calibration-free sensing system, characterized in that, The system includes a data integration module, a compensation module, a reliability verification module, and a parameter update module. The device is an electronic device with data processing capabilities, used to implement the sensor system environmental compensation method as described in any one of claims 1 to 9. The data integration module receives raw measurement data and real-time environmental monitoring data, uses a preset data alignment algorithm to temporally correlate the raw measurement data and real-time environmental monitoring data, and converts them into synchronous measurement data. It then constructs a measurement matrix X and an environmental matrix E based on the synchronous measurement data. The compensation module calculates the true value vector μ, the influence coefficient matrix β, and the compensated measurement value matrix. The reliability verification module is used to perform reliability verification; the parameter update module is used to optimize and update the influence coefficient matrix β when the parameter update is triggered.

11. The device according to claim 10, characterized in that, The device is an embedded processor or an application-specific integrated circuit, wherein the embedded processor includes an ARM processor or an FPGA processor.

12. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the sensing system environmental compensation method as described in any one of claims 1 to 9.

13. A measurement system for environmental compensation of a calibration-free sensing system, characterized in that, The system includes a measurement sensing device, a calculation verification unit, and an environmental variable monitoring device. The measurement sensing device includes at least one sensor for acquiring raw measurement data; the measurement sensing device and the calculation verification unit are data-connected. The environmental variable monitoring device includes at least one sensor for acquiring real-time environmental monitoring data; the environmental variable monitoring device and the calculation verification unit are data-connected. Both the measurement sensing device and the environmental variable monitoring device complete m synchronous measurements within a preset time window. The calculation verification unit includes a calculation compensation module and a verification evaluation module. The calculation compensation module is used to calculate intermediate parameters without calibration based on the data obtained from the m synchronous measurements, and to compensate for the measured values ​​based on the intermediate parameters. The verification evaluation module is used to verify and evaluate the intermediate parameters; if the verification evaluation fails, it triggers an update of the intermediate parameters. The calculation verification unit is used to implement the environmental compensation method for the sensing system as described in any one of claims 1 to 9.

14. The measurement system according to claim 13, characterized in that, The computational verification unit is either an embedded processor or an application-specific integrated circuit.

15. The measurement system according to claim 13, characterized in that, The computational verification unit is implemented by an executable program in an edge computing device or a cloud server cluster.

16. The measurement system according to claim 13, characterized in that, The verification and evaluation module uses at least one of the statistical error index, confidence index, distribution characteristic index, and dynamic performance index to verify and evaluate the intermediate parameters.

Images