Temperature compensation method and system, detection equipment, computer equipment and storage medium

Abstract

The invention provides a temperature compensation method and system, detection equipment, computer equipment and a storage medium. The method comprises the following steps: acquiring measurement data of a target element after hardware compensation; processing the measurement data by adopting a pre-calibrated Monte Carlo incremental model, and calculating to obtain a residual drift value of each sample track at the current step length; reconstructing real output data based on each residual drift value to obtain a candidate output set; fusing the candidate output sets by adopting a preset decision strategy to obtain prediction information; a temperature drift compensation amount is generated based on the prediction information. According to the method, most of linear temperature excursions are eliminated in a hardware compensation mode, on this basis, the Monte Carlo incremental model is further adopted, the temperature excursion track is accurately predicted at each step length, real-time compensation is carried out on the target element according to the predicted temperature excursion track, a small part of remaining nonlinear temperature excursions are eliminated, and the accuracy of the temperature excursion is improved. The temperature compensation effect of the target element is improved, and high-precision measurement in a wide temperature range changing environment is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of semiconductor detection, and in particular to a temperature compensation method and system, a detection device, a computer device and a storage medium. BACKGROUND

[0002] In the process of semiconductor thin film epitaxial growth, the wafer surface temperature field is crucial to nucleation kinetics and epitaxial rate, which directly affects the growth rate, film quality, doping efficiency, component ratio and interface steepness. The epitaxial growth process has a wide temperature range of about 500°C - 1200°C, and the temperature uniformity has a great influence on the device performance. Multi-zone heating disc needs to cooperate with real-time in-situ detection device for accurate and stable detection. In the traditional technology, a combined detection system (Reflectance-Temperature-Curvature, RTC) is usually used to realize the above-mentioned in-situ online detection function. The components in the combined detection system generally have the problem of electrical parameter drift with temperature change, which directly affects the measurement accuracy and system stability.

[0003] Negative temperature coefficient thermistor (NTC) is widely used in temperature sensing and compensation link due to its low cost and high sensitivity. However, the material and structural characteristics of NTC determine that there is thermal equilibrium hysteresis: when the environment temperature changes step by step or rapidly, the heat transfer process cannot be completed instantaneously, and the response of the device temperature to the environment temperature shows obvious time constant and hysteresis; superimposed with the inconsistency of device self-heating effect (caused by bias power consumption) and packaging / installation thermal resistance, a persistent time sequence deviation is generated between the temperature measurement value and the actual temperature of the compensated target.

[0004] Traditional compensation technology focuses on static error correction, which relies on fixed parameter mapping relationship and cannot represent the coupling effect of temperature change rate and response hysteresis, and has fundamental limitations in adapting to the accurate detection of wide temperature range changes in the epitaxial growth process. SUMMARY

[0005] The present application aims to provide a temperature compensation method, system, detection device, computer device and storage medium to overcome the limitation of traditional compensation technology that relies on fixed parameter mapping relationship and cannot adapt to the accurate detection of wide temperature range changes in the process of thin film epitaxial growth.

[0006] In a first aspect, the present application provides a temperature compensation method, comprising: obtaining measurement data of a target element at a preset step size; wherein the measurement data comprises temperature data and output voltage compensated by a hardware compensation unit. The pre-calibrated Monte Carlo incremental model is used to calculate residual drift values of each sample trajectory at a current step based on the measurement data, wherein each sample trajectory is generated by the Monte Carlo incremental model based on the measurement data; Real output data is reconstructed based on the residual drift values of each current step to obtain a candidate output set; A preset decision strategy is used to fuse the candidate output set to obtain prediction information; A temperature drift compensation amount is generated based on the prediction information, so that the hardware compensation unit compensates the target element for temperature according to the temperature drift compensation amount.

[0007] In one of the embodiments, the hardware compensation unit includes two operational amplifiers of symmetric design, and the temperature drift characteristics of the two operational amplifiers are matched. In one of the embodiments, the way of calibrating the Monte Carlo incremental model includes: Sample data of the target element is obtained at intervals of a preset sampling period, wherein the sample data includes temperature data, output voltage and reference voltage; the output voltage is obtained after the target element is compensated by hardware; Based on the output voltage and the reference voltage, a residual drift voltage is calculated; Based on the residual drift voltage and the temperature data, a temperature drift characteristic parameter is constructed; wherein the temperature drift characteristic parameter includes at least one of temperature change rate, temperature change rate change speed, residual error integral and residual error absolute value; Based on the temperature drift characteristic parameter, a mean function and a standard deviation function of residual drift increment are constructed to obtain the Monte Carlo incremental model.

[0008] In one of the embodiments, the mean function includes a temperature change compensation coefficient; and the standard deviation function includes an aging coefficient and a temperature disturbance gain; The way of calibrating the Monte Carlo incremental model further includes: A linear regression relationship between residual error and temperature change rate square is measured under the condition of rapid temperature cycling to obtain the temperature change compensation coefficient; An aging test is performed at a constant temperature change rate to establish a linear relationship between running time and measured standard deviation function to obtain the aging coefficient; The standard deviation of residual error is measured at different temperature change rates, and the temperature disturbance gain is determined according to the slope of the amplitude of the temperature change rate.

[0009] In one of the embodiments, the pre-calibrated Monte Carlo incremental model is used to calculate residual drift values of each sample trajectory at a current step based on the measurement data, including: The Monte Carlo incremental model is used to calculate a mean value and a standard deviation value corresponding to the measurement data of the current step; random disturbance is introduced on the basis of the mean value and the standard deviation value to obtain a residual drift increment of the current step; The residual drift increment of each sample trajectory at the current step is superimposed on the residual drift value of the previous step to obtain a residual drift value of each sample trajectory at the current step.

[0010] In one of the embodiments, the number of sample trajectories is dynamically set, and the number is positively related to the absolute value of the temperature change rate in a hyperbolic tangent relationship.

[0011] In one of the embodiments, the reconstruction of the real output data based on the residual drift values of the current steps to obtain a candidate output set includes: For each sample trajectory, the output voltage of the target element is subtracted from the residual drift value of the sample trajectory at the current step to obtain real output data; Each of the real output data constitutes the candidate output set.

[0012] In one of the embodiments, the adoption of a preset decision strategy to fuse the candidate output set to obtain prediction information includes: Based on statistical distance measurement, the consistency of each candidate value in the candidate output set and its corresponding sample trajectory with the set distribution is evaluated, and abnormal trajectories and candidate values whose corresponding candidate outputs significantly deviate from the group distribution are identified and excluded to obtain an effective candidate set; According to the similarity between the current temperature change rate and the temperature change rate of each sample trajectory, weights are assigned to the effective candidate set to obtain a weighted candidate set; The weighted candidate set is subjected to quantile aggregation to generate point estimates and confidence intervals of the current step to obtain the prediction information.

[0013] In one of the embodiments, the generation of a temperature drift compensation amount based on the prediction information includes: According to the point estimate, an initial compensation amount of the current step is generated; Based on the confidence interval, the initial compensation amount is adaptively processed to obtain a temperature drift compensation amount.

[0014] In one of the embodiments, the method further includes: Based on the temperature data, a nonlinear characteristic parameter is constructed; wherein the nonlinear characteristic parameter includes at least one of a temperature change rate, a change amount of the temperature change rate, and a time decay term of temperature change intensity; Based on an adaptive weight, each of the nonlinear characteristic parameters is weighted and summed to obtain a fine-tuning compensation amount; Based on the fine adjustment compensation amount, the initial compensation amount after adaptive processing is feedforward corrected to obtain a temperature drift compensation amount.

[0015] In one of the embodiments, the method further comprises: generating calibration instructions according to the prediction information to adjust parameters of the hardware compensation unit; and / or adjusting parameters of the Monte Carlo increment model according to the temperature compensation effect of the target element.

[0016] In one of the embodiments, the method further comprises: calculating a temperature change rate according to the temperature data; comparing the temperature change rate with a preset threshold, and switching the working mode according to the comparison result; wherein the working mode comprises an energy-saving mode, a high-energy mode, and a standard mode.

[0017] In a second aspect, the present application provides a temperature compensation system, which comprises: an acquisition module configured to acquire measurement data of a target element at preset intervals; wherein the measurement data comprises temperature data and output voltage after hardware compensation; a prediction module configured to calculate residual drift values of each sample trajectory at a current step based on the measurement data by using a pre-calibrated Monte Carlo increment model; wherein each sample trajectory is generated by the Monte Carlo increment model based on the measurement data; reconstruct real output data based on the residual drift values of each current step to obtain a candidate output set; and fuse the candidate output set by using a preset decision strategy to obtain prediction information; a compensation module configured to generate a temperature drift compensation amount based on the prediction information, so that the hardware compensation unit performs temperature compensation on the target element according to the temperature drift compensation amount.

[0018] In a third aspect, the present application further provides a detection device, which comprises: an acquisition module configured to acquire temperature information of a wafer when performing a thin film epitaxial growth process; a temperature detection module configured to monitor temperature changes of the wafer in the thin film growth process according to the temperature information; the temperature detection module comprises a target element and a hardware compensation unit; The temperature compensation system in the second aspect is configured to acquire measurement data of the target element, and generate a temperature drift compensation amount based on the measurement data, so that the hardware compensation unit performs temperature compensation on the target element according to the temperature drift compensation amount.

[0019] In a fourth aspect, the present application provides a computer device. The computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method steps in the first aspect.

[0020] In a fifth aspect, the present application provides a computer readable storage medium. The computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method steps in the first aspect.

[0021] The temperature compensation method, system, detection device, computer device and storage medium have at least the following advantages: The present application obtains the measurement data of the target element after hardware compensation, processes the measurement data by using the pre-calibrated Monte Carlo incremental model, calculates the residual drift values of each sample trajectory at the current step, reconstructs the real output data based on the residual drift values, obtains a candidate output set, fuses the candidate output set by using a preset decision strategy, obtains prediction information, generates a temperature drift compensation amount based on the prediction information, and compensates the target element for temperature. The present application firstly eliminates most linear temperature drift by using the hardware compensation method, further uses the Monte Carlo incremental model to accurately predict the temperature drift trajectory at each step, and then compensates the target element in real time according to the predicted temperature drift trajectory, thereby eliminating the remaining small part of nonlinear temperature drift, improving the temperature compensation effect of the target element, and realizing high-precision measurement in a wide temperature range. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 FIG. 1 is a structural block diagram of a detection device in one embodiment; Figure 2 FIG. 2 is a flowchart of a temperature compensation method in one embodiment; Figure 3 FIG. 3 is a structural schematic diagram of a hardware compensation unit in one embodiment; Figure 4 FIG. 4 is a structural block diagram of a hardware compensation unit in another embodiment; Figure 5 FIG. 5 is a flowchart of the step of calibrating the Monte Carlo incremental model in one embodiment; Figure 6 FIG. 6 is a flowchart of the step of calculating the residual drift value in one embodiment; Figure 7 FIG. 7 is a flowchart of the step of fusing the candidate output set to obtain the prediction information in one embodiment; Figure 8 FIG. 8 is a structural block diagram of a temperature compensation system in one embodiment; Figure 9 FIG. 9 is an internal structural diagram of a computer device in one embodiment. DETAILED DESCRIPTION

[0023] The present application is described and explained with additional specificity and detail through the use of the accompanying drawings in which:

[0024] Some example embodiments of the present application are described for illustrative purposes with reference to accompanying drawings, it being understood that the present application can be carried out in other ways not specifically shown in the drawings.

[0025] Referring to Figure 1 , Figure 1 The process cavity 1 in the above-mentioned device is provided with a machine table tray 11 and a wafer 12, and the wafer 12 is placed on the machine table tray 11. Optionally, in one embodiment, the present application provides a detection device for detecting the temperature change of the wafer 12 during thin film growth in an epitaxial growth process. The detection device comprises a collection module 2, a temperature detection module 3 and a temperature compensation system 4.

[0026] The collection module 2 is connected to the temperature detection module 3, and is used to obtain temperature information of the wafer 12 when the thin film epitaxial growth process is performed, and transmit the temperature information to the temperature detection module 3. It should be understood that the collection module 2 in the present embodiment is used to obtain temperature information, and in actual application, the collection module 2 can also obtain reflectivity and / or curvature as needed. Illustratively, the collection module 2 comprises a probe, a transmitting optical fiber and a collection optical fiber; wherein the probe is used to coaxially integrate the optical path of the transmitting optical fiber and the collection optical fiber and to perform point and angle coupling on the wafer surface, to limit the measurement spot size and working distance; the transmitting optical fiber is used to introduce the detection light source into the cavity and to irradiate the wafer surface. The collection optical fiber is used to collect thermal radiation and reflection signals from the wafer surface and transmit them to the temperature detection module 3.

[0027] The temperature detection module 3 is connected to the temperature compensation system 4, and is used to monitor the temperature change of the wafer 12 during thin film growth according to the temperature information. The temperature detection module 3 comprises a target element 31 and a hardware compensation unit 32. Wherein the target element 31 is one component in the hardware detection link of the temperature detection module 3, and the component may

[0028] The temperature compensation system 4 is configured to obtain measurement data of the target element 31, and generate a temperature drift compensation amount based on the measurement data, so that the hardware compensation unit 32 compensates the target element 31 according to the temperature drift compensation amount. The measurement data includes temperature data and output voltage compensated by the hardware compensation unit 32.

[0029] Specifically, the temperature compensation system 4 outputs the temperature drift compensation amount according to the measurement data, including: The temperature compensation system 4 obtains the measurement data of the target element at intervals of a preset step length. The measurement data includes temperature data and output voltage compensated by the hardware compensation unit. A pre-calibrated Monte Carlo incremental model is used to calculate the residual drift values of each sample trajectory at the current step length based on the measurement data. Each sample trajectory is generated by the Monte Carlo incremental model based on the measurement data. The real output data is reconstructed based on the residual drift values of each current step length to obtain a candidate output set. A preset decision strategy is used to fuse the candidate output set to obtain prediction information. The temperature drift compensation amount is generated based on the prediction information.

[0030] The above detection device, the temperature detection module monitors the temperature change of the wafer in the thin film growth process according to the temperature information obtained by the acquisition module; the temperature compensation system obtains the measurement data of the target element at intervals of a preset step length, and processes the measurement data by using a pre-calibrated Monte Carlo incremental model to calculate the residual drift values of each sample trajectory at the current step length; the real output data is reconstructed based on each residual drift value to obtain a candidate output set; a preset decision strategy is used to fuse the candidate output set to obtain prediction information; the temperature drift compensation amount is generated based on the prediction information; and the hardware compensation unit compensates the target element based on the temperature drift compensation amount. The present application first uses a hardware compensation method to eliminate most of the linear temperature drift, and then uses a Monte Carlo incremental model to accurately predict the temperature drift trajectory at each step length, and then compensates the target element in real time according to the predicted temperature drift trajectory, thereby eliminating the remaining small part of the nonlinear temperature drift, improving the temperature compensation effect of the target element, and realizing high-precision measurement of the temperature detection module in a wide temperature range change environment.

[0031] In one exemplary embodiment, the present embodiment provides a temperature compensation method, which is applied to the server 104 in Figure 1 as an example for description.

[0032] Please refer to Figure 2 , Figure 2 is a flowchart of a temperature compensation method of the present embodiment, which specifically includes the following steps: Step 202, obtaining the measurement data of the target element at intervals of a preset step length. The measurement data includes temperature data and output voltage compensated by the hardware compensation unit.

[0033] Specifically, the preset step length is a discrete time interval for temperature drift recursion and data sampling. In the embodiment, the preset step length refers to the time required for completing the entire process of data measurement, model inference, and trajectory update. Exemplarily, the preset step length Δt can be set according to the sampling frequency, and the expression is ; wherein, is the sampling frequency, which is determined by the system clock or the sampling rate of the analog-to-digital conversion module, to ensure that the model inference is aligned with the sampling. It should be noted that the value of the preset step length Δt should be greater than or equal to the minimum stable value allowed by the measurement link bandwidth, so as to capture the rapid temperature change of the target element on the premise of ensuring that the computing resources are acceptable.

[0034] In step 204, the residual drift value of each sample trajectory in the current step length is calculated based on the measurement data by using the pre-calibrated Monte Carlo incremental model; wherein each sample trajectory is generated by the Monte Carlo incremental model based on the measurement data; the real output data is reconstructed based on the residual drift value of each current step length to obtain a candidate output set; and the candidate output set is fused by using a preset decision strategy to obtain prediction information.

[0035] Specifically, the Monte Carlo method is a numerical calculation method for solving mathematical, physical and engineering problems by using random sampling and statistics. In the embodiment, the Monte Carlo incremental model gradually predicts the residual temperature drift of the target element during the working process, calculates the increment by introducing random disturbance in each preset step length, and then obtains the residual drift value of the current step length.

[0036] The sample trajectory refers to that, in each step length, the measurement data corresponding to the step length is processed by using the Monte Carlo incremental model to obtain a processing result, and the random disturbance is independently sampled multiple times based on the processing result.

[0037] In step 206, a temperature drift compensation amount is generated based on the prediction information, so that the hardware compensation unit compensates the temperature of the target element according to the temperature drift compensation amount.

[0038] The above-mentioned temperature compensation method obtains the measurement data of the target element after hardware compensation; uses a pre-calibrated Monte Carlo incremental model to process the measurement data, and calculates the residual drift value of each sample trajectory at the current step length; reconstructs the real output data based on each residual drift value to obtain a candidate output set; uses a preset decision strategy to fuse the candidate output set to obtain prediction information; generates a temperature drift compensation amount based on the prediction information to perform temperature compensation on the target element. This application first uses hardware compensation to eliminate most of the linear temperature drift, and on this basis further uses a Monte Carlo incremental model to accurately predict the temperature drift trajectory at each step length, and then performs real-time compensation on the target element according to the predicted temperature drift trajectory, eliminating the remaining small part of the nonlinear temperature drift, improving the temperature compensation effect of the target element, and achieving high-precision measurement in a wide temperature range environment.

[0039] See also Figure 3 , Figure 3 Figure 2 shows the schematic diagram of the hardware compensation unit. The hardware compensation unit includes two symmetrically designed operational amplifiers with matching temperature drift characteristics.

[0040] Specifically, the hardware compensation unit includes two operational amplifiers (hereinafter referred to as the main operational amplifier U1 and the compensation operational amplifier U2). The temperature drift characteristics of the main operational amplifier U1 and the compensation operational amplifier U2 are matched, and the main operational amplifier U1 and the compensation operational amplifier U2 are configured as Figure 3 In the differential input, single-ended output structure shown, the useful signal in this circuit is the difference between the input and output of the main op amp U1 and the compensation op amp U2. In other embodiments, the main op amp U1 and the compensation op amp U2 can also be configured with a differential input, differential output structure as needed.

[0041] Using the above solution, the ambient temperature change of the target component ( Or a slow drift over time (t). Under ideal matching, the drift voltage / current generated by the main op amp U1 and the compensation op amp U2 should be highly correlated, with very similar amplitudes and directions. This results in these two highly correlated drift components being largely canceled out as common-mode signals at the output of the hardware compensation unit. Ultimately, the residual temperature drift at the output is much smaller than the drift of a single op amp.

[0042] Optionally, the expression of the temperature drift voltage transfer function of the target component is:

[0043] in, Indicates the temperature drift coefficient; represents the time drift coefficient; represents the temperature change acceleration compensation term; This transfer function describes the temperature drift voltage that remains after cancellation and cannot be completely eliminated. It consists of three parts, corresponding to different physical mechanisms of drift.

[0044] Linear temperature drift term: Even if the main operational amplifier U1 and the compensation operational amplifier U2 are strictly screened and matched, the parameters of their core devices cannot be absolutely identical in terms of temperature variation characteristics. This slight mismatch will cause the above-mentioned linear temperature drift term, which is the main source of residual temperature drift.

[0045] Time drift term: Semiconductor devices and internal materials will experience extremely slow aging effects over time, causing their bias voltage and current to drift. Even if two operational amplifiers are perfectly matched at the initial time, there may be slight differences in their aging rates.

[0046] Temperature variation acceleration compensation term: When the ambient temperature changes rapidly, there may be transient temperature differences in different positions inside the operational amplifier chip. Even if the package is tightly adjacent, heat conduction still needs time. This transient spatial temperature gradient will cause the main operational amplifier U1 and the compensation operational amplifier U2 to experience slight differences in temperature T1(t) and T2(t) at the same time. This transient mismatch drift caused by thermal gradient is usually proportional to the square of the temperature change rate dT / dt.

[0047] Generally, ambient temperature compensation can be divided into the following two scenarios.

[0048] In the temperature rise scenario, the output of the main operational amplifier U1 adds a positive temperature drift voltage +ΔVt to its useful signal. Since the main operational amplifier U1 and the compensation operational amplifier U2 have highly matched temperature drift characteristic parameters and are in the same thermal environment, the main operational amplifier U1 simultaneously generates a positive compensation voltage +V't at its output. The compensation voltage +V't is injected into the inverting input of the compensation operational amplifier U2, which injects a positive voltage (+V't) into the inverting input of the compensation operational amplifier U2. According to the principle of virtual short and virtual open at the inverting input of the operational amplifier and the characteristics of the inverting amplifier, this will cause the output of the compensation operational amplifier U2 to produce a negative change -G·V't, where G is the gain of the compensation operational amplifier U2 from the injection point to the output. The main operational amplifier U1 temperature drift signal +ΔVt reaches the inverting terminal of the compensation operational amplifier U2, generating a reverse compensation current -Icomp. The output change of the compensation operational amplifier U2 can be represented as: ΔV_out = -G·(ΔVt - V't) ≈ 0.

[0049] In the temperature drop scenario, the output of the main operational amplifier U1 adds a negative temperature drift voltage -ΔVt to the useful signal, and for the same reason, the compensation operational amplifier U2 generates a negative compensation voltage -V't, which is injected into the inverting input of the compensation operational amplifier U2 to generate a reverse compensation current +Icomp, and the output change of the compensation operational amplifier U2 can be expressed as: ΔV_out = G·(ΔVt - V't)≈ 0 By using the above scheme, the hardware compensation unit is connected to the output of the target element, and the hardware compensation is performed, so that most of the linear temperature drift can be eliminated. However, since the two temperature coefficients cannot be completely matched, the final output voltage signal includes a residual temperature drift voltage, which is denoted as residual drift voltage. The expression of the residual drift voltage is:

[0050] wherein, , ΔVt and V't respectively represent the actual temperature drift coefficients of the main operational amplifier U1 and the compensation operational amplifier U2; ΔT represents the change of the ambient temperature.

[0051] The expression of the final output voltage signal is:

[0052] wherein, S(t) is the target voltage corresponding to the reference temperature at time t.

[0053] The voltage signal is the output voltage of the target element after the hardware compensation.

[0054] Please refer to Figure 4 Optionally, the hardware compensation unit further comprises a temperature sensor and a compensation digital-to-analog converter.

[0055] The temperature sensor is arranged close to the target element, and is used to detect the temperature data of the target element and transmit the temperature data to the temperature compensation system. The temperature compensation system is used to calculate the temperature compensation value according to the measurement data of the target element.

[0056] The compensation digital-to-analog converter is connected to the temperature compensation system and the compensation operational amplifier U2 respectively, and is used to convert the digital temperature compensation value into a controllable analog current, and inject the analog current into the input of the compensation operational amplifier U2 to offset the residual temperature drift in the output voltage of the target element.

[0057] Exemplarily, the expression of the compensation current injected into the input of the compensation operational amplifier U2 is:

[0058]

[0059] wherein, represents a compensation coefficient; represents an equivalent conversion coefficient of the compensation current on the output; represents an output voltage of the input voltage Vin after the main operational amplifier U1 and the compensation operational amplifier U2.

[0060] Please refer to Figure 5 Optionally, the manner of calibrating the Monte Carlo incremental model comprises: In step 502, sample data of the target element is obtained at intervals of a preset sampling period. The sample data comprises temperature data, an output voltage and a reference voltage. The output voltage is obtained after hardware compensation of the target element.

[0061] In step 504, a residual drift voltage is calculated based on the output voltage and the reference voltage.

[0062] In step 506, a temperature drift characteristic parameter is constructed based on the residual drift voltage and the temperature data. The temperature drift characteristic parameter comprises at least one of a temperature change rate, a change speed of the temperature change rate, a residual error integral quantity and a residual error absolute value.

[0063] In step 508, a mean function and a standard deviation function of the residual drift increment are constructed based on the temperature drift characteristic parameter, to obtain a Monte Carlo incremental model.

[0064] Specifically, the preset sampling period is set in the same manner as the preset step length, and the value of the preset sampling period can be the same as or different from the preset step length.

[0065] In each preset sampling period, current sample data of the target element is obtained.

[0066] The temperature gradient is calculated in real time according to the temperature data, and the expression is:

[0067] wherein, k represents the current time; represents a sampling period, and Δt = 1 / fsample.

[0068] The residual drift voltage is calculated according to the output voltage and the reference voltage, and the expression is:

[0069] wherein, represents a target voltage corresponding to the current temperature.

[0070] The expression of the temperature drift characteristic parameter is: X = [ , , , . represents the temperature change rate, Rate of change of temperature change, Residual integral, Residual absolute value, And Susceptible to circuit noise pollution, can be through the Kalman filtering, Using moving average filter processing, and filter window ≈ 10x thermal time constant; When the cumulative amount exceeds the threshold, the early warning device is aged and fails; Abnormal oscillation, then prompt the heat dissipation system failure.

[0071] Through real-time feature extraction and adaptive compensation closed loop, the application can maintain sub-microvolt level stability in temperature dramatic change, long-term aging and other scenes, and provide core guarantee for high-precision measurement system.

[0072] Further, the embodiments of the application adopt mean function and standard deviation function to jointly construct the framework of the Monte Carlo incremental model, wherein the mean function and the standard deviation function are functions of the temperature gradient ∇T, and the mean function is composed of inherent drift coefficient and residual temperature drift coefficient from hardware design, and the standard deviation function is jointly determined by initial random disturbance and gain term growing with aging, and is updated with sampling length. The above parameters jointly constitute the model parameters of the Monte Carlo incremental model.

[0073] The expression of the Monte Carlo incremental model is:

[0074] Wherein, The mean function is represented by; The standard deviation function is represented by; Random number between 0 and 1 is represented by.

[0075]

[0076] Wherein, Temperature change compensation coefficient from hardware design is represented by, Residual temperature drift coefficient is represented by.

[0077]

[0078] Wherein, Initial random disturbance is represented by, Aging coefficient is represented by, Temperature disturbance gain is represented by.

[0079] Optionally, in the case that the mean function includes the temperature change compensation coefficient; the standard deviation function includes the aging coefficient and the temperature disturbance gain, the way of calibrating the Monte Carlo incremental model further includes: The linear regression relationship between the residual error and the square of the temperature change rate is measured under the condition of rapid temperature cycling to obtain a temperature change compensation coefficient; The aging test is performed under a constant temperature change rate to establish a linear relationship between the running time and the measured standard deviation function to obtain an aging coefficient; The standard deviation of the residual error is measured under different temperature change rates, and the temperature disturbance gain is determined according to the slope of the standard deviation with respect to the amplitude of the temperature change rate.

[0080] Specifically, in addition to providing the above function equation, the calibration of the Monte Carlo incremental model also needs to determine the parameters of each function to obtain a specific Monte Carlo incremental model.

[0081] For example, when calibrating the temperature change compensation coefficient The rapid temperature cycling is performed, for example, under the condition of 5-20℃ / min, and the linear regression relationship between the residual error and the square of the temperature change rate is measured, which is expressed as:

[0082] Wherein, cov(·,·) represents the covariance; var(·) represents the variance.

[0083] When calibrating the temperature disturbance gain The standard deviation of the residual error is measured under different temperature change rates, which is expressed as:

[0084] By using the above scheme, the temperature change compensation coefficient is obtained through rapid temperature cycling, so that the mean function is timely corrected with respect to the temperature change rate; the linear relationship between the running time and the standard deviation function is established through the aging test to obtain the aging coefficient, so that the fluctuation intensity of the standard deviation function is sensitive to aging and does not drift; the standard deviation of the residual error is measured under different temperature change rates to determine the temperature disturbance gain, so that the variance characterization realizes the self-adaptation to the temperature slowly changing or rapidly changing scene, and the baseline offset and the systematic drift term are determined under the static stable working condition. The Monte Carlo incremental model formed in this way makes the random increment statistics of each sampling step accurately match the current temperature change, and the Monte Carlo sample trajectory not only covers the real uncertainty but also avoids divergence. Compared with the traditional static compensation, it can be operated for a long time and maintain high precision and robustness.

[0085] Please refer to Figure 6 Alternatively, the pre-calibrated Monte Carlo incremental model is used to calculate the residual drift value of each sample trajectory at the current step based on the measured data, including: Step 602, using the Monte Carlo incremental model, calculating the mean and standard deviation values corresponding to the measured data at the current step; introducing random disturbance based on the mean and standard deviation values to obtain the residual drift increment at the current step.

[0086] Step 604, superimpose the residual drift increment of each sample trajectory at the current step length to the residual drift value at the last step length to obtain the residual drift value of each sample trajectory at the current step length.

[0087] Specifically, at each preset step length, measurement data is obtained, and the measurement data is input into the mean function and the standard deviation function to calculate the corresponding mean and standard deviation. On this basis, a random number Zi is randomly assigned to each sample trajectory, thereby obtaining a plurality of different increment samples.

[0088] For the ith sample trajectory, the expression of the residual drift increment is:

[0089] wherein, denotes the discrete time step length, for ensuring the variance scale invariance.

[0090] Superimpose the above residual drift increment on the residual drift value calculated at the last step length to obtain the residual drift increment at the current step length, and the expression is:

[0091] Repeat the above process N times until the calculation of all samples is completed, and obtain a sample set of , which represents the residual temperature drift trajectory cluster that may occur under the current characteristic condition.

[0092] Optionally, the real output data is reconstructed based on the residual drift value of each current step length to obtain a candidate output set, including: For each sample trajectory, the output voltage of the target element is subtracted from the residual drift value of the sample trajectory at the current step length to obtain the real output data; and each real output data constitutes a candidate output set.

[0093] Specifically, for the ith sample trajectory, a virtual signal is constructed:

[0094] That is, the actual output voltage is subtracted from the above residual drift value to generate N possible prediction distributions, which strips the influence of temperature drift, thereby obtaining the real path that the output voltage may exist.

[0095] Optionally, the number of sample trajectories is dynamically set, and the number and the absolute value of the temperature change rate are in a positive correlation hyperbolic tangent relationship.

[0096] Specifically, the expression of the mapping relationship between the number of sample trajectories and the temperature change rate is:

[0097] wherein, represents the minimum sampling number in steady state, represents the maximum sampling number when temperature changes dramatically, represents the sampling density growth coefficient.

[0098] By using the above scheme, the number of sample trajectories is adjusted in real time according to the temperature change rate, and is constrained by the minimum sampling number and the maximum sampling number, so that the calculation power consumption can be reduced and the inference speed of the model can be improved in the temperature slowly changing scene, and the real-time performance is met. Meanwhile, the number of sample trajectories is self-matched according to the temperature change scene, over-compensation is avoided, the interval reliability is improved, and the inference accuracy is further improved.

[0099] Please refer to Figure 7 Optionally, the candidate output set is fused by using a preset decision strategy to obtain prediction information, including: In step 702, based on the statistical distance metric, the consistency of each candidate value in the candidate output set and its corresponding sample trajectory with the set distribution is evaluated, and abnormal trajectories and candidate values whose corresponding candidate outputs significantly deviate from the group distribution are identified and excluded, to obtain an effective candidate set.

[0100] In step 704, according to the similarity between the current temperature change rate and the temperature change rate of each sample trajectory, weights are assigned to the effective candidate set to obtain a weighted candidate set.

[0101] In step 706, the weighted candidate set is subjected to quantile aggregation to generate point estimates and confidence intervals of the current step length, and prediction information is obtained.

[0102] Specifically, in order to extract the most reliable prediction information from a large number of random trajectories output by the Monte Carlo incremental model, the present application combines three intelligent mechanisms of abnormal elimination, trajectory weighting and quantile aggregation to further fuse the candidate output set.

[0103] The abnormal elimination strategy is used to identify and exclude abnormal trajectories deviating from the group based on the statistical distance metric to obtain an effective candidate set.

[0104] The trajectory weighting strategy is used to assign weights to each trajectory in the effective candidate set, and to preferentially select the trajectory most similar to the current temperature dynamics. The Euclidean distance between the weight of each trajectory and the current temperature change rate and the trajectory temperature change rate is in negative exponential relationship, and the weight expression is:

[0105] wherein, represents the current measured temperature change rate, represents the temperature change rate of trajectory i, and λ is the weight attenuation coefficient.

[0106] Quantile aggregation strategy is used to fuse each sample trajectory to obtain optimal prediction information. The prediction value at a specific probability quantile is selected, and the weighted fusion is performed according to the inverse ratio of the probability product. The expression is as follows:

[0107]

[0108] wherein, represents the point estimate obtained by fusion; represents the candidate output set; represents the q quantile; represents the weight, which is inversely proportional to the probability product , which is used to appropriately strengthen the two end quantiles.

[0109] Further, the upper and lower confidence quantiles p1 and p2 are selected, wherein p1 < p2; the confidence interval of the current step is given as follows:

[0110] By using the above scheme, the point estimate is used to calculate and apply the temperature compensation amount within the current step, and the confidence interval is used to evaluate the estimation uncertainty and perform adaptive gain adjustment on the temperature compensation amount, so as to suppress overcompensation or undercompensation and improve compensation accuracy. The above decision strategy converts the probability prediction into an executable decision, so that the temperature drift compensation system has the industrial intelligent characteristics of self-evaluation, self-optimization and self-warning, and provides a core guarantee for high-reliability electronic systems.

[0111] Optionally, the temperature drift compensation amount is generated based on the prediction information, including: An initial compensation amount of the current step is generated according to the point estimate; and the initial compensation amount is adaptively processed based on the confidence interval to obtain the temperature drift compensation amount.

[0112] Specifically, the direction and amplitude of compensation are determined according to the point estimate to form an initial compensation amount; and whether the current step needs compensation and the size of the compensation are determined according to the confidence of the confidence interval, so as to finally obtain the temperature drift compensation amount. The temperature drift compensation amount is written into a compensation digital-to-analog converter and converted into an analog current signal, which is injected into the input end of the compensation operational amplifier U2 to offset the residual temperature drift in the output voltage of the target element.

[0113] Optionally, in an exemplary embodiment, the above temperature compensation method further includes: Based on the temperature data, a nonlinear characteristic parameter is constructed; wherein the nonlinear characteristic parameter includes at least one of a temperature change rate, a change amount of the temperature change rate, and a time decay term of the temperature change intensity; Based on the adaptive weight, the weighted sum of each nonlinear characteristic parameter is obtained to obtain a fine adjustment compensation amount; The fine adjustment compensation amount is used to feed forward correct the initial compensation amount after adaptive processing to obtain the temperature drift compensation amount.

[0114] Specifically, the fine adjustment compensation amount is calculated based on a non-linear compensation model, and the expression is as follows:

[0115] wherein, represents the fine adjustment compensation amount; represents the temperature change rate; represents the change of the temperature change rate; represents the temperature change intensity; represents the exponential decay memory, representing that the short-term inertia or aging effect after mutation will decay over time, is a time constant; , , represents the adaptive weight.

[0116] At each preset step, the error between the estimation of the residual drift of the non-linear compensation model and the estimation of the residual drift of the Monte Carlo incremental model is compared, and the adaptive weight and the time constant are adjusted according to the error weighted by the confidence for small step updating. The adjusted adaptive weight and time constant will be used in the calculation of the next preset step.

[0117] By using the above scheme, the result of the Monte Carlo incremental model is used as the main compensation amount, and the fine adjustment compensation amount is used to superimpose it to reduce the residual error and jitter, thereby improving the prediction accuracy.

[0118] Optionally, in an exemplary embodiment, the temperature compensation method further comprises: generating calibration instructions according to the prediction information to adjust the parameters of the hardware compensation unit; and / or adjusting the parameters of the Monte Carlo incremental model according to the temperature compensation effect of the target element.

[0119] Specifically, the embodiment of the present application realizes the leap of temperature drift compensation from the traditional single-layer suppression to the multi-level collaborative optimization based on the bidirectional coupling mechanism of the hardware layer and the software layer, and realizes high-precision measurement through the deep collaboration of hardware and software.

[0120] wherein, the forward learning process refers to that the probability model generates hardware calibration instructions to adjust the parameters of the hardware compensation unit. Exemplarily, the parameters of the hardware compensation unit include compensation coefficients According to the prediction information, the calibration instructions are generated to adjust the parameters of the hardware compensation unit, including: estimating the slope of the residual drift to the temperature, if the slope is greater than 0, it indicates that the output increases with the temperature, and the compensation coefficient needs to be reduced; otherwise, it is increased, so as to fine tune the compensation coefficient accordingly.

[0121] The reverse adaptation process is to adjust the probability model parameters by detecting the hardware parameter drift, and to adjust the probability model in reverse to compensate for the hardware drift.

[0122] Optionally, in an exemplary embodiment, the temperature compensation method further comprises: According to the temperature data, the temperature change rate is calculated; the temperature change rate is compared with the preset threshold, and the working mode is switched according to the comparison result; wherein the working mode includes energy saving mode, high energy mode, standard mode.

[0123] Specifically, the energy saving mode is applicable to the case where the environmental temperature changes little, and the activation condition is that when the temperature change rate is lower than the preset threshold, and the duration reaches the set value, or the system idle time reaches the set value.

[0124] The high energy mode is applicable to the case where the temperature changes rapidly, at this time the system increases the sampling rate, enables the high frequency algorithm, and runs at ultra-high frequency.

[0125] The standard mode is the default start mode, and the switching condition from the energy saving mode is that the temperature gradient is greater than the preset threshold; and the switching condition from the high energy mode is that the temperature change rate reaches the threshold, and the duration reaches the set value.

[0126] In order to realize the intelligent switching of the energy saving mode, the standard mode and the high energy mode, an intelligent switching controller is introduced, which intelligently switches according to the threshold switching condition, realizes the dynamic optimization of the temperature drift compensation system in multiple dimensions such as precision, power consumption and response speed through bidirectional learning and intelligent mode switching, and provides the core ability of self-adapting to environmental changes for various electronic systems.

[0127] It should be understood that although each step in the flowchart involved in each embodiment as described above is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other order. Moreover, at least part of the steps in the flowchart involved in each embodiment as described above can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or steps or stages in other steps.

[0128] Based on the same inventive concept, the embodiments of the present application also provide a temperature compensation system, which is applicable to the above-mentioned temperature compensation method, and the implementation scheme for solving the problem provided by the system is similar to the implementation scheme described in the above method. Therefore, the specific limitations in one or more device embodiments provided below can refer to the limitations of the method described above, and will not be described here.

[0129] Referring to Figure 8 In one embodiment, the temperature compensation system comprises an acquisition module, a prediction module and a compensation module.

[0130] The acquisition module is configured to acquire measurement data of the target element at preset intervals; wherein the measurement data comprises temperature data and output voltage after hardware compensation.

[0131] The prediction module is configured to calculate residual drift values of each sample trajectory at a current step based on the measurement data by using a pre-calibrated Monte Carlo incremental model; wherein each sample trajectory is generated by the Monte Carlo incremental model based on the measurement data; reconstruct real output data based on the residual drift values at each current step to obtain a candidate output set; and fuse the candidate output set by using a preset decision strategy to obtain prediction information.

[0132] The compensation module is configured to generate a temperature drift compensation amount based on the prediction information, so that the hardware compensation unit performs temperature compensation on the target element according to the temperature drift compensation amount.

[0133] Optionally, the temperature compensation system further comprises a calibration module.

[0134] The calibration module is configured to calibrate the Monte Carlo incremental model, and the calibration manner comprises: acquiring sampling data of the target element at preset sampling periods. Wherein the sampling data comprises temperature data, output voltage and reference voltage; the output voltage is obtained after hardware compensation of the target element; residual drift voltage is calculated based on the output voltage and the reference voltage; temperature drift characteristic parameters are constructed based on the residual drift voltage and the temperature data; wherein the temperature drift characteristic parameters comprise at least one of a temperature change rate, a change speed of the temperature change rate, a residual error integral quantity and a residual error absolute value; a mean function and a standard deviation function of residual drift increment are constructed based on the temperature drift characteristic parameters to obtain the Monte Carlo incremental model.

[0135] Optionally, in the case that the mean function comprises a temperature variation compensation coefficient; and the standard deviation function comprises an aging coefficient and a temperature disturbance gain, the calibration manner of the calibration module for calibrating the Monte Carlo incremental model further comprises: measuring a linear regression relationship between the residual error and the square of the temperature change rate under a rapid temperature cycling condition to obtain the temperature variation compensation coefficient; performing an aging test under a constant temperature change rate to establish a linear relationship between the running time and the measured standard deviation function to obtain the aging coefficient; and measuring the standard deviation of the residual error under different temperature change rates, and determining the temperature disturbance gain according to the slope of the standard deviation with respect to the amplitude of the temperature change rate.

[0136] Optionally, the prediction module employs a pre-calibrated Monte Carlo incremental model to calculate the residual drift value of each sample trajectory at the current step based on the measurement data, including: employing the Monte Carlo incremental model to calculate the mean value and standard deviation value corresponding to the measurement data at the current step; introducing random disturbance based on the mean value and standard deviation value to obtain the residual drift increment at the current step; superimposing the residual drift increment of each sample trajectory at the current step on the residual drift value at the previous step to obtain the residual drift value of each sample trajectory at the current step. The number of sample trajectories is dynamically set, and the number is positively related to the absolute value of the temperature change rate in a hyperbolic tangent relationship.

[0137] Optionally, the prediction module reconstructs the real output data based on the residual drift value of each current step to obtain a candidate output set, including: for each sample trajectory, subtracting the output voltage of the target element from the residual drift value of the sample trajectory at the current step to obtain the real output data; each real output data constitutes the candidate output set.

[0138] Optionally, the prediction module employs a pre-set decision strategy to fuse the candidate output set to obtain the prediction information, including: based on the statistical distance measure, evaluating the consistency of each candidate value in the candidate output set and its corresponding sample trajectory with the set distribution, identifying and excluding abnormal trajectories and candidate values whose corresponding candidate outputs deviate significantly from the group distribution to obtain an effective candidate set; according to the similarity between the current temperature change rate and the temperature change rate of each sample trajectory, assigning weights to the effective candidate set to obtain a weighted candidate set; performing quantile aggregation on the weighted candidate set to generate point estimates and confidence intervals at the current step to obtain the prediction information.

[0139] Optionally, the compensation module generates a temperature drift compensation amount based on the prediction information, including: generating an initial compensation amount at the current step according to the point estimate; based on the confidence interval, adaptively processing the initial compensation amount to obtain the temperature drift compensation amount.

[0140] Optionally, the above-mentioned temperature compensation system further comprises: a nonlinear compensation module.

[0141] The nonlinear compensation module is configured to construct nonlinear characteristic parameters based on the temperature data; wherein the nonlinear characteristic parameters include at least one of the temperature change rate, the change amount of the temperature change rate, and the time decay term of the temperature change intensity; perform weighted summation on each nonlinear characteristic parameter based on the adaptive weight to obtain a fine tuning compensation amount; perform feedforward correction on the initial compensation amount that has been adaptively processed based on the fine tuning compensation amount to obtain the temperature drift compensation amount.

[0142] Optionally, the above-mentioned temperature compensation system further comprises: a calibration module.

[0143] The calibration module is used to generate calibration instructions based on the prediction information to adjust the parameters of the hardware compensation unit; and / or adjust the parameters of the Monte Carlo incremental model based on the temperature compensation effect of the target component.

[0144] Optionally, the temperature compensation system further includes: a working mode switching module.

[0145] The working mode switching module is used to calculate the temperature change rate based on the temperature data; compare the temperature change rate with the preset threshold, and switch the working mode according to the comparison result; among which, the working modes include energy-saving mode, high-energy mode, and standard mode.

[0146] The above-mentioned temperature compensation system obtains the measurement data of the target element after hardware compensation; uses a pre-calibrated Monte Carlo incremental model to process the measurement data, and calculates the residual drift value of each sample trajectory at the current step length; reconstructs the real output data based on each residual drift value to obtain a candidate output set; uses a preset decision strategy to fuse the candidate output set to obtain prediction information; generates a temperature drift compensation amount based on the prediction information to perform temperature compensation on the target element. This application first uses hardware compensation to eliminate most of the linear temperature drift, and on this basis further uses a Monte Carlo incremental model to accurately predict the temperature drift trajectory at each step length, and then compensates the target element in real time according to the predicted temperature drift trajectory, eliminating the remaining small part of the nonlinear temperature drift, improving the temperature compensation effect of the target element, and achieving high-precision measurement in a wide temperature range environment. Each module in the temperature compensation system described above may be implemented in whole or in part through software, hardware, or a combination thereof. Each module may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0147] In a feasible embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 9The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to perform wired or wireless communication with external terminals. The wireless communication can be achieved through WIFI, mobile cellular network, NFC (Near Field Communication) or other technologies. The computer program is executed by the processor to implement the temperature compensation method. The display unit of the computer device is configured to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0148] Those skilled in the art can understand that, Figure 9 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0149] In one possible implementation, a computer device is provided, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method steps of the temperature compensation method.

[0150] In one possible implementation, a computer readable storage medium is provided, which stores a computer program. The computer program is executed by a processor to implement the method steps of the temperature compensation method.

[0151] In one possible implementation, a computer program product is provided, which includes a computer program. The computer program is executed by a processor to implement the method steps of the temperature compensation method.

[0152] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0153] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.

[0154] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A temperature compensation method, characterized in that: The method comprises: Acquire measurement data of the target element at intervals of a preset step size; wherein the measurement data includes temperature data and output voltage compensated by a hardware compensation unit; Using a pre-calibrated Monte Carlo incremental model, the residual drift value of each sample trajectory at the current step length is calculated based on the measurement data; wherein each sample trajectory is generated by the Monte Carlo incremental model based on the measurement data; Reconstructing the real output data based on the residual drift values ​​of each current step length to obtain a candidate output set; Using a preset decision strategy to fuse the candidate output sets to obtain prediction information; A temperature drift compensation amount is generated based on the prediction information, so that the hardware compensation unit performs temperature compensation on the target element according to the temperature drift compensation amount.

2. The method according to claim 1, characterized in that The hardware compensation unit includes two symmetrically designed operational amplifiers, and the temperature drift characteristics of the two operational amplifiers are matched.

3. The method according to claim 1, characterized in that Methods for calibrating Monte Carlo incremental models include: Acquire sampling data of a target element at intervals of a preset sampling period, wherein the sampling data includes temperature data, output voltage, and reference voltage; the output voltage is obtained after hardware compensation is performed on the target element; Calculating a residual drift voltage based on the output voltage and the reference voltage; constructing a temperature drift characteristic parameter based on the residual drift voltage and the temperature data; wherein the temperature drift characteristic parameter includes at least one of a temperature change rate, a change speed of the temperature change rate, a residual integral, and a residual absolute value; Based on the temperature drift characteristic parameters, a mean function and a standard deviation function of the residual drift increment are constructed to obtain the Monte Carlo increment model.

4. The method according to claim 3, characterized in that The mean function includes a temperature variation compensation coefficient; the standard deviation function includes an aging coefficient and a temperature disturbance gain; Methods for calibrating the Monte Carlo incremental model also include: The linear regression relationship between the residual error and the square of the temperature change rate is measured under rapid temperature cycling conditions to obtain the temperature change compensation coefficient; An aging test is performed at a constant temperature change rate to establish a linear relationship between the operating time and the measured standard deviation function to obtain the aging coefficient; The standard deviation of the residual is measured at different temperature change rates, and the temperature disturbance gain is determined based on its slope with the amplitude of the temperature change rate.

5. The method according to claim 1, wherein The method of using a pre-calibrated Monte Carlo incremental model to calculate the residual drift value of each sample trajectory at the current step length based on the measurement data includes: The Monte Carlo incremental model is used to calculate the mean and standard deviation corresponding to the measurement data of the current step length; random disturbance is introduced based on the mean and the standard deviation to obtain the residual drift increment of the current step length; The residual drift increment of each sample trajectory in the current step length is added to the residual drift value of the previous step length to obtain the residual drift value of each sample trajectory in the current step length.

6. The method according to claim 5, characterized in that The number of the sample trajectories is dynamically set, and the number is positively correlated with the absolute value of the temperature change rate in a hyperbolic tangent relationship.

7. The method according to claim 1, characterized in that The actual output data is reconstructed based on the residual drift values ​​of each current step length to obtain a candidate output set, including: For each sample trajectory, the output voltage of the target element is subtracted from the residual drift value of the sample trajectory at the current step length to obtain the actual output data; The actual output data constitute the candidate output set.

8. The method according to claim 1, characterized in that The adopting of a preset decision strategy to fuse the candidate output set to obtain prediction information includes: Based on the statistical distance metric, the consistency of each candidate value and its corresponding sample trajectory in the candidate output set with the set distribution is evaluated, and abnormal trajectories and candidate values ​​whose corresponding candidate outputs significantly deviate from the group distribution are identified and excluded to obtain a valid candidate set; Assigning weights to the valid candidate set according to the similarity between the current temperature change rate and the temperature change rate of each sample trajectory to obtain a weighted candidate set; Quantile aggregation is performed on the weighted candidate set to generate a point estimate and a credible interval of the current step size to obtain the prediction information.

9. The method according to claim 8, characterized in that Generating a temperature drift compensation amount based on the prediction information includes: generating an initial compensation amount for the current step length based on the point estimate; Based on the credible interval, the initial compensation amount is adaptively processed to obtain a temperature drift compensation amount.

10. The method according to claim 9, characterized in that The method further comprises: Constructing a nonlinear characteristic parameter based on the temperature data; wherein the nonlinear characteristic parameter includes at least one of a temperature change rate, a change in the temperature change rate, and a time decay term of the temperature change intensity; Based on the adaptive weights, weighted summation is performed on each of the nonlinear characteristic parameters to obtain a fine-tuning compensation amount; Based on the fine adjustment compensation amount, a feedforward correction is performed on the initial compensation amount that has undergone adaptive processing to obtain a temperature drift compensation amount.

11. The method according to claim 1, characterized in that The method further comprises: generating a calibration instruction according to the prediction information to adjust parameters of the hardware compensation unit; and / or adjusting parameters of the Monte Carlo incremental model according to the temperature compensation effect of the target element.

12. The method according to claim 1, characterized in that The method further comprises: Calculating a temperature change rate based on the temperature data; The temperature change rate is compared with a preset threshold, and the working mode is switched according to the comparison result; wherein the working mode includes energy-saving mode, high-energy mode, and standard mode.

13. A temperature compensation system, characterized in that: The system comprises: An acquisition module, configured to acquire measurement data of a target element at intervals of a preset step size; wherein the measurement data includes temperature data and an output voltage compensated by a hardware compensation unit; a prediction module, using a pre-calibrated Monte Carlo incremental model to calculate a residual drift value of each sample trajectory at a current step length based on the measurement data; wherein each sample trajectory is generated by the Monte Carlo incremental model based on the measurement data; reconstructing true output data based on the residual drift value of each current step length to obtain a candidate output set; and fusing the candidate output set using a preset decision strategy to obtain prediction information; The compensation module is configured to generate a temperature drift compensation amount based on the prediction information, so that the hardware compensation unit performs temperature compensation on the target element according to the temperature drift compensation amount.

14. A detection device, characterized in that: include: An acquisition module is used to obtain temperature information of the wafer when performing a thin film epitaxial growth process; A temperature detection module, configured to monitor temperature changes of the wafer during thin film growth based on the temperature information; the temperature detection module includes a target element and a hardware compensation unit; The temperature compensation system as described in claim 13 is used to obtain measurement data of the target element and generate a temperature drift compensation amount based on the measurement data, so that the hardware compensation unit performs temperature compensation on the target element according to the temperature drift compensation amount.

15. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 12 are implemented.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

Images