ARC test temperature rise rate calculation method and device

Abstract

The invention provides an ARC test temperature rise rate calculation method and device, and belongs to the technical field of ARC test. The ARC test temperature rise rate calculation method comprises the following steps: S1, obtaining original time temperature data of an ARC test search stage; s2, smoothing the original time and temperature data based on a non-parametric smoothing method to obtain a central trend curve, and scanning the central trend curve based on a change point detection algorithm to extract a plurality of data segments; s3, fitting all the data segments based on a robust regression algorithm to obtain a target linear model; and S4, obtaining the slope of the target linear model, and taking the slope as the temperature rise rate of the ARC test search stage. According to the method, the technical problem that the calculation result is inaccurate and unstable due to data noise and abnormal values when the temperature rise rate at the Search stage of the ARC test is calculated by adopting an end point method in the prior art is solved, the accuracy is improved, the robustness and the repeatability are relatively high, the hardware dependence is reduced, and automatic and intelligent analysis is realized.

Description

Technical Field

[0001] This invention relates to the field of ARC testing technology, and more specifically, to a method and apparatus for calculating the temperature rise rate in ARC testing. Background Technology

[0002] Adiabatic accelerated calorimetry (ARC) is a key instrument for evaluating the thermal stability of substances such as chemicals and battery materials. Its standard testing procedure typically follows a "heat-wait-search" pattern. The "search" phase aims to detect the initial, weak heat release of the sample with extremely high sensitivity. The temperature rise rate (dT / dt) in this phase is the core evaluation indicator, directly determining the onset temperature and influencing the final assessment of the sample's safety performance.

[0003] Currently, the common data processing method in the industry is the endpoint method: directly extract the temperature values ​​at the start and end times of each "Search" stage, and calculate the average temperature rise rate by (T_{end}-T_{start}) / (t_{end}-t_{start}). Where T is temperature and t represents time. This method has the following significant drawbacks: (1) Sensitive to noise: ARC test data are generally affected by sensor noise, ambient temperature fluctuations and system electromagnetic interference, which manifests as high-frequency noise. When the instantaneous temperature value at the start or end point happens to fall on the peak or trough of the noise, it will cause a huge deviation in the calculation results, which cannot truly reflect the physical process; (2) Lack of resistance to outliers: During the experiment, occasional instantaneous system disturbances will produce isolated outlier data points. If these outliers happen to appear at the endpoints of the interval, the calculation results will be completely distorted; (3) Poor repeatability: Due to the above sensitivity to noise and outliers, even repeated tests of the same sample may yield significantly different temperature rise rates using the endpoint method, affecting the stability of the results and the comparability between different experiments; (4) Ignoring nonlinear dynamics: The real exothermic process has parts of generation and conduction, and is often not an ideal straight line, but often exhibits S-shaped or other complex nonlinear dynamic characteristics. Using only a single linear model to summarize the entire process will inevitably introduce systematic bias; (5) Inability to handle structured pseudo-signals: In addition to random noise, ARC data often contains structured pseudo-signals caused by physicochemical processes, which are manifested as short-term temperature drops or plateaus. Such non-random "outlier segments" will seriously mislead traditional regression algorithms. In addition, the endpoint method can only obtain a single average temperature rise rate, and cannot meet the needs of deeper data analysis, such as identifying and quantifying the maximum temperature rise rate corresponding to the maximum instantaneous risk. Summary of the Invention

[0004] To address the aforementioned issues, this application provides a method and apparatus for calculating the temperature rise rate in ARC testing.

[0005] Firstly, this application provides a real-time example of a method for calculating the temperature rise rate in an ARC test, comprising the following steps:

[0006] S1: Obtain the raw time and temperature data for the ARC test search phase;

[0007] S2: The original time-temperature data is smoothed using a non-parametric smoothing method to obtain a central trend curve. The central trend curve is then scanned using a change point detection algorithm to extract multiple data segments, each of which includes a series of data points.

[0008] S3: Fit all data segments using a robust regression algorithm to obtain the target linear model;

[0009] S4: Obtain the slope of the target linear model and use this slope as the temperature rise rate in the ARC test search phase.

[0010] Preferably, the nonparametric smoothing method in step S2 is the LOESS algorithm, and the robust regression algorithm in step S3 is the RANSAC algorithm.

[0011] Preferably, the method of smoothing the original time-temperature data based on the LOESS algorithm to obtain a central trend curve specifically includes: setting a local window, wherein the number of data points in the local window accounts for a preset proportion of the total number of data points; sliding the local window along a preset direction to traverse the entire original time-temperature data; performing weighted linear regression on multiple original data points located in the window each time using the LOESS algorithm to obtain multiple smoothed data points; and generating a smooth central trend curve based on the multiple smoothed data points.

[0012] Preferably, step S3 specifically includes:

[0013] S31: Set the tolerance threshold for each data segment;

[0014] S32: Based on the tolerance threshold, iteratively calculate each data segment to obtain the point set within the corresponding data segment;

[0015] S33: Combine all the point sets within the data segments to form the optimal internal point set;

[0016] S34: Fit the data points in the optimal interior point set using the least squares method to obtain the target linear model.

[0017] Preferably, step S31 specifically includes:

[0018] S311: Calculate the absolute deviation of the median for each data segment;

[0019] S312: Calculate the tolerance threshold corresponding to the data segment by multiplying the median absolute deviation by a preset constant.

[0020] Preferably, step S32 specifically includes:

[0021] S321: Randomly select a subset from the data segment and construct the current linear model based on the selected subset;

[0022] S322: Calculate the distance from all data points in the acquired data segment to the current linear model, and mark the data points whose distance is less than the tolerance threshold as the current inner point set;

[0023] S323: Obtain the historical inner point set corresponding to the data segment. If there is no historical inner point set, use the current inner point set as the historical inner point set and the current linear model as the optimal model. If the number of points in the current inner point set is greater than the number of points in the historical inner point set, use the current inner point set as the historical inner point set and the current linear model as the optimal model.

[0024] S324: Return to S321 for iterative calculation until the number of iterations is the preset value, and use the historical point set as the point set of the corresponding data segment.

[0025] Preferably, step S1 specifically includes: loading an ARC test data file; and filtering the original time and temperature data of the ARC test search stage from the ARC test data file according to the stage identifier.

[0026] Secondly, embodiments of this application provide an ARC test temperature rise rate calculation device, including a data acquisition module, a smoothing processing module, a robust regression module, and a slope acquisition module;

[0027] The data acquisition module is used to acquire the raw time and temperature data during the ARC test search phase;

[0028] The smoothing module is used to smooth the original time-temperature data based on a non-parametric smoothing method to obtain a central trend curve, and to scan the central trend curve based on a change point detection algorithm to extract multiple data segments, each of which includes a series of data points.

[0029] The robust regression module is used to fit all data segments based on the robust regression algorithm to obtain the target linear model;

[0030] The slope acquisition module is used to obtain the slope of the target linear model and use the slope as the temperature rise rate in the ARC test search phase.

[0031] Preferably, the nonparametric smoothing method is the LOESS algorithm; the smoothing module performs smoothing processing on the original time-temperature data based on the nonparametric smoothing method to obtain the central trend curve, specifically including: setting a local window, wherein the number of data points in the local window accounts for a preset proportion of the total number of data points; sliding the local window along a preset direction to traverse the entire original time-temperature data, and performing weighted linear regression on multiple original data points located in the window each time using the LOESS algorithm to obtain multiple smoothed data points; and generating a smoothed central trend curve based on the multiple smoothed data points.

[0032] Preferably, the robust regression algorithm is the RANSAC algorithm; the robust regression module includes a threshold setting unit, an iterative calculation unit, an interior point set unit, and a fitting unit;

[0033] The threshold setting unit is used to set the tolerance threshold for each data segment respectively;

[0034] The iterative calculation unit is used to perform iterative calculations on each data segment based on the tolerance threshold to obtain the data segment intra-point set corresponding to the data segment.

[0035] The inner point set unit is used to combine all the inner point sets of the data segments to form the optimal inner point set;

[0036] The fitting unit is used to fit the data points in the optimal interior point set based on the least squares method to obtain the target linear model.

[0037] The beneficial effects of this invention are as follows: It organically links nonparametric smoothing methods and robust regression algorithms. First, nonparametric smoothing methods capture local nonlinear trends in data and filter out high-frequency noise. Then, robust regression algorithms eliminate structured pseudo-signals and outlier interference. The two form a complementary synergistic mechanism, realizing full-process optimization of "noise filtering - pseudo-signal identification - robust fitting". This solves the technical problem in the prior art where the calculation of the temperature rise rate in the "Search" stage of ARC testing using the endpoint method is inaccurate and unstable due to data noise and outliers. It improves accuracy, has strong robustness and repeatability, reduces hardware dependence, and realizes automated and intelligent analysis. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 A flowchart illustrating an ARC test temperature rise rate calculation method provided in this application embodiment;

[0040] Figure 2 This is a schematic diagram of the structure of an ARC test temperature rise rate calculation device provided in an embodiment of this application;

[0041] Figure 3 This is a comparison chart of the fitting results obtained in thermal analysis using the endpoint method, the RANSAC algorithm alone, and a combination of the LOESS and RANSAC algorithms. Detailed Implementation

[0042] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0043] In the following description, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. The following description provides multiple embodiments of this application, which can be substituted or combined with each other. Therefore, this application can also be considered to include all possible combinations of the same and / or different embodiments described. Thus, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then this application should also be considered to include embodiments containing one or more other possible combinations of A, B, C, and D, even if such embodiments are not explicitly described in the following text.

[0044] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes may be made to the function and arrangement of the described elements without departing from the scope of this application. Various processes or components may be appropriately omitted, substituted, or added to the examples. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Furthermore, features described with respect to some examples may be combined into other examples.

[0045] Please see Figure 1 . Figure 1 This is a flowchart illustrating a method for calculating the temperature rise rate in an ARC test according to an embodiment of this application. In this embodiment, the method includes the following steps:

[0046] S1: Obtain the raw time and temperature data for the ARC test search phase;

[0047] S2: The original time-temperature data is smoothed using a non-parametric smoothing method to obtain a central trend curve. The central trend curve is then scanned using a change point detection algorithm to extract multiple data segments, each of which includes a series of data points.

[0048] S3: Fit all data segments using a robust regression algorithm to obtain the target linear model;

[0049] S4: Obtain the slope of the target linear model and use this slope as the temperature rise rate in the ARC test search phase.

[0050] In this embodiment, the original time-temperature data is smoothed using a nonparametric smoothing method to solve the problems of nonlinear dynamics and high-frequency noise. A robust regression algorithm is used to fit all data segments to solve the problems of pseudo-signals and outlier interference and to achieve fitting. This can overcome the interference of noise, outliers and pseudo-signals, and calculate the temperature rise rate of the "Search" stage more accurately and robustly.

[0051] In this application, we present an experimental data processing method for evaluating the thermal stability of materials, which can be used for thermal analysis. This application can be used to accurately and robustly calculate the temperature rise rate (dT / dt) during the "Search" phase of an Accelerated Adiabatic Calorimeter (ARC) test. This application provides an automated, high-precision, and highly robust data processing method to obtain a temperature rise rate that better represents the actual physical process. It solves the technical problem in existing technologies where the endpoint method is used to calculate the temperature rise rate during the "Search" phase of the ARC test, leading to inaccurate and unstable calculation results due to data noise and outliers.

[0052] In this application embodiment, the application calculates the ARC test temperature rise rate based on a nonparametric smoothing method and a robust regression algorithm. Instead of relying on two endpoints in isolation, it utilizes all data points within the "Search" phase to extract the main trend of temperature change by establishing a target linear model.

[0053] In this embodiment, data smoothing and segmentation are performed in step S2. The original time-temperature data is the entire time-temperature data of the ARC test search phase. A non-parametric smoothing method is used to smooth the entire time-temperature data of the ARC test search phase. A change point detection algorithm automatically identifies and extracts the data segments to be analyzed. Each data segment contains a series of (time t_i, temperature T_i) data points. In step S3, linear model fitting is performed. For each data segment, a robust regression algorithm is applied to fit and establish a target linear model T = a*t + b, where T is temperature and t is time. In step S3, slope extraction and result output are performed. The slope 'a' is extracted from the fitted target linear model T = a*t + b, and the extracted slope 'a' is used as the optimized temperature rise rate (dT / dt) for this "Search" phase for subsequent thermal stability analysis. This application organically combines the non-parametric smoothing method with the robust regression algorithm, enabling them to work synergistically and achieving a "1+1 > 2" technical effect.

[0054] In the embodiments of this application, each data segment includes a series of data points, indicating that each data segment includes multiple data points.

[0055] In one possible implementation, the nonparametric smoothing method in step S2 is the LOESS algorithm, and the robust regression algorithm in step S3 is the RANSAC algorithm.

[0056] In one possible implementation, smoothing the original time-temperature data based on the LOESS algorithm to obtain a central trend curve specifically includes: setting a local window, wherein the number of data points within the local window accounts for a preset proportion of the total number of data points; sliding the local window along a preset direction to traverse the entire original time-temperature data; performing weighted linear regression on multiple original data points within the window each time using the LOESS algorithm to obtain multiple smoothed data points; and generating a smooth central trend curve based on the multiple smoothed data points.

[0057] In this embodiment, the nonparametric smoothing method in step S2 is the LOESS algorithm. Therefore, the process of smoothing the original time-temperature data using the nonparametric smoothing method to obtain the central trend curve in step S2 is equivalent to smoothing the original time-temperature data using the LOESS algorithm to obtain the central trend curve. Using the LOESS algorithm to smooth the original time-temperature data to obtain the central trend curve solves the problems of nonlinear dynamics and high-frequency noise. As a local regression smoother, LOESS does not pre-define a global model and can flexibly capture the local nonlinear trends of the data. The preset ratio can be 10%, meaning that the number of data points within the local window accounts for 10% of the total number of data points. The LOESS algorithm performs weighted linear regression on the original data points within this local window to obtain multiple smoothed data points corresponding to this portion of the original data points. Each time the local window is slid, the original data points within the local window are updated once. The LOESS algorithm then performs weighted linear regression on the original data points within the local window to obtain multiple smoothed data points corresponding to this portion of the original data points. This process continues until the entire original time-temperature data is traversed by sliding the local window, and multiple smoothed data points corresponding to all original data points are obtained. Based on these multiple smoothed data points, a smooth central trend curve is generated. The central trend curve can truly reflect the macroscopic outline of the exothermic process and effectively filter out high-frequency random noise.

[0058] In this embodiment of the application, step S2 involves scanning the central trend curve based on a change point detection algorithm to extract multiple data segments. Each data segment includes a series of data points. Specifically, this includes calling a change point detection algorithm (e.g., an algorithm based on abrupt changes in data slope and noise level) to scan the entire data sequence, automatically identifying and segmenting all independent temperature change stages, and extracting them into independent data segments D = {(t1, T1), (t2, T2), ..., (t_n, T_n)}.

[0059] In this embodiment, the purpose of step S2 is to effectively filter out high-frequency random noise and initially fit the local nonlinear characteristics of the data, so as to provide a "cleaner" input for subsequent signal screening and robust regression.

[0060] In the embodiments of this application, see Figure 3 , Figure 3 This is a comparison chart of the fitting results obtained in thermal analysis using the endpoint method, the RANSAC algorithm alone, and the combination of the LOESS and RANSAC algorithms. The thin line segment represents the fitting result obtained using only the RANSAC algorithm, and the thick line segment represents the fitting result obtained using the combination of the LOESS and RANSAC algorithms.

[0061] In one possible implementation, step S3 specifically includes:

[0062] S31: Set the tolerance threshold for each data segment;

[0063] S32: Based on the tolerance threshold, iteratively calculate each data segment to obtain the point set within the corresponding data segment;

[0064] S33: Combine all the point sets within the data segments to form the optimal internal point set;

[0065] S34: Fit the data points in the optimal interior point set using the least squares method to obtain the target linear model.

[0066] In this embodiment, the robust regression algorithm in step S3 is the RANSAC algorithm. Step S3 involves fitting all data segments to obtain the target linear model based on the RANSAC algorithm. For data segments D with consistent trends after LOESS smoothing, the RANSAC (Random Sample Consensus) algorithm is applied to fit the linear model, which can solve the problems of pseudo-signals and outlier interference and achieve fitting. For data segments with consistent trends after LOESS smoothing, the RANSAC (Random Sample Consensus) robust regression algorithm is applied. The RANSAC algorithm finds "interior points" in the data through iteration, that is, data that conforms to the trend of the main model, and constructs the final linear model only based on these interior points. This mechanism can ignore isolated outliers and structured pseudo-signals (such as plateau periods or sudden drops) because these data segments cannot form enough consistent samples, thus being identified as "outside points" and excluded from the final calculation.

[0067] In one possible implementation, step S31 specifically includes:

[0068] S311: Calculate the absolute deviation of the median for each data segment;

[0069] S312: Calculate the tolerance threshold corresponding to the data segment by multiplying the median absolute deviation by a preset constant.

[0070] In this embodiment, step S31 performs adaptive parameter setting: First, the median absolute deviation (MAD) of data segment D is calculated. MAD = median (|T_i - median (T)|). Then, the tolerance threshold ε of RANSAC is automatically set based on the MAD value (e.g., ε = 2.5MAD). This allows the algorithm to adaptively determine whether a point is an interior point based on the noise level of the current data, and then use it for fitting. The median absolute deviation (MAD) of the data segment is calculated as MAD = median (|T_i - median (T)|). median: median, refers to the value in the middle position after arranging a set of data in ascending order. If the amount of data is odd, the median value is taken; if it is even, the average of the two middle numbers is taken; |Ti - median (T)|: represents the absolute deviation (i.e., absolute distance) between each data point Ti and the median of the dataset T; median (|Ti - median (T)|) is the median of all absolute deviations, i.e., "median of absolute deviation". The RANSAC tolerance threshold ε is automatically set based on the median absolute deviation, where ε = k * MAD, k is a preset constant (e.g., 2.5), and MAD represents the median absolute deviation. This allows the algorithm to adaptively determine whether a point is an inlier based on the noise level of the current data. The median absolute deviation is a discrete metric insensitive to outliers; using it to set the threshold allows the algorithm to adapt to different data noise levels in different experiments and stages, thus achieving automated processing with minimal human intervention.

[0071] In one possible implementation, step S32 specifically includes:

[0072] S321: Randomly select a subset from the data segment and construct the current linear model based on the selected subset;

[0073] S322: Calculate the distance from all data points in the acquired data segment to the current linear model, and mark the data points whose distance is less than the tolerance threshold as the current inner point set;

[0074] S323: Obtain the historical inner point set corresponding to the data segment. If there is no historical inner point set, use the current inner point set as the historical inner point set and the current linear model as the optimal model. If the number of points in the current inner point set is greater than the number of points in the historical inner point set, use the current inner point set as the historical inner point set and the current linear model as the optimal model.

[0075] S324: Return to S321 for iterative calculation until the number of iterations is the preset value, and use the historical point set as the point set of the corresponding data segment.

[0076] In this embodiment, step S32 iteratively searches for the optimal model, and the algorithm enters an iterative loop. In each iteration, a minimum subset (2 points for a linear model) is randomly selected from the data to construct a temporary linear model M_temp. Then, the distance from all data points to this model is calculated, and points whose distance is less than the tolerance threshold ε are marked as the current inlier set I_temp. The size of the current inlier set I_temp is compared with the size of the historical inlier set I_best, that is, the number of points in the current inlier set is compared with the number of points in the historical inlier set. If |I_temp|>|I_best|, then I_best=I_temp is updated. That is, if the number of points in the current inlier set is greater than the number of points in the historical inlier set, then the current inlier set is used as the historical inlier set, and the current linear model M_temp is recorded as the optimal model M_best. This iterative process is repeated a specified number of times, such as 100 times. 100 times is a robust choice, and the process of the algorithm finding the optimal fitting parameters can be completed faster. The preset value can be 100. When the number of iterations is a preset value, the historical intrapoint set is used as the intrapoint set of the corresponding data segment.

[0077] In this embodiment of the application, for each data segment, the method of step S32 is used to obtain the data segment intra-point set corresponding to the data segment, and the data segment intra-point sets corresponding to all data segments are combined to form the optimal intra-point set.

[0078] In this embodiment, step S34 involves final model fitting. Using all data points in the optimal interior set I_best, a final and most robust linear model T=at+b is fitted using the standard least squares method.

[0079] In this embodiment, step S4 involves extracting and outputting the temperature rise rate. The slope parameter 'a' is directly extracted from the final fitted target linear model T = a*t + b. This slope 'a' represents the optimized temperature rise rate (dT / dt) that indicates the main heating trend of the "Search" stage. The calculated temperature rise rate 'a', along with the corresponding "Search" stage number, start time, and other information, is output to the results file for subsequent thermodynamic analysis, heat release initiation temperature determination, and other uses.

[0080] In one possible implementation, step S1 specifically includes: loading an ARC test data file; and filtering the raw time and temperature data of the ARC test search phase from the ARC test data file according to the phase identifier. The complete ARC test data file includes data corresponding to each phase of the ARC test, and the raw time and temperature data of the ARC test search phase is filtered from the ARC test data file according to the phase identifier.

[0081] In this embodiment of the application, a complete ARC test data file is loaded through a data interface. This file contains a sequence of timestamps (t) and corresponding temperature values ​​(T) sampled at fixed time intervals.

[0082] In this application embodiment, a "smooth-robust regression" collaborative processing architecture is constructed, which breaks through the inherent limitations of the traditional endpoint method, specifically in the following aspects:

[0083] (1) Collaborative algorithm architecture design: The LOESS nonparametric smoothing and RANSAC robust regression are organically linked. First, the LOESS algorithm captures the local nonlinear trend of the data and filters out high-frequency noise, providing a high-quality data foundation for subsequent regression; then, the RANSAC algorithm eliminates structured pseudo-signals and outlier interference. The two form a complementary collaborative mechanism to achieve full-process optimization of "noise filtering - pseudo-signal identification - robust fitting", which solves the technical problem that a single algorithm cannot handle noise, nonlinearity and outliers at the same time.

[0084] (2) Dynamic parameter adaptive mechanism: An adaptive parameter adjustment strategy based on data characteristics is proposed. In the RANSAC algorithm, the tolerance threshold ε=k*MAD is dynamically set by calculating the median absolute deviation (MAD) of the data segment. This enables the algorithm to automatically adjust the judgment criteria according to the noise level of different experimental data, avoiding the problem of poor adaptability of traditional fixed thresholds to data with different noise intensities. This significantly improves the versatility and automation of the algorithm.

[0085] (3) Phased data processing strategy: In view of the coexistence of nonlinearity and overall linear trend in the data of the ARC “Search” phase, a phased processing approach of “first nonlinear smoothing and then linear fitting” is adopted. The LOESS algorithm does not pre-define a global model and can flexibly adapt to complex nonlinear dynamic characteristics such as S-shape, preserving the macro trend of the data; the subsequent RANSAC linear fitting is based on the purified data trend, taking into account the needs of local nonlinear description and overall linear quantization, and overcoming the modeling bias of the traditional endpoint method for nonlinear processes;

[0086] (4) Full data segment modeling instead of endpoint sampling: Abandoning the traditional endpoint method that relies on only two endpoint data points, modeling and analysis are performed using all data points in the "Search" stage. By using sliding window weighted regression and iterative filtering of internal points, the effective information in the data is fully explored, the impact of individual outliers on the results is reduced, and the calculated temperature rise rate can better reflect the true heat release trend of the entire stage, which greatly improves the stability and accuracy of the results;

[0087] (5) Automated Process Integration: The system integrates functional modules such as change point detection, LOESS smoothing, and RANSAC fitting of dynamic parameters into an automated processing flow, achieving unattended operation from data loading to result output. The algorithm automatically completes data segmentation, noise filtering, outlier identification, and model fitting, avoiding subjective errors caused by manual intervention and providing technical support for high-throughput ARC test data analysis.

[0088] In the embodiments of this application, compared with the prior art (endpoint method), this application has the following significant beneficial effects:

[0089] (1) Significantly improved accuracy: By robustly fitting all data points in the “Search” stage, this method effectively averages the impact of high-frequency noise and avoids calculation errors caused by the randomness of endpoint data;

[0090] (2) Strong robustness and repeatability: The RANSAC algorithm used in the preferred scheme can automatically identify and eliminate interference from outlier data points and structured pseudo-signals, making the analysis results insensitive to fluctuations in data quality. This greatly improves the consistency and repeatability of data analysis results between different batches and different samples;

[0091] (3) Achieving automated and intelligent analysis: Through automatic segmentation using the change point detection algorithm and dynamic setting of RANSAC parameters using MAD, the entire data processing flow can run fully automatically. This lays the technical foundation for establishing a high-throughput ARC data analysis platform and greatly improves analysis efficiency;

[0092] (4) The physical meaning is clearer and the analysis dimensions are richer: The slope obtained by fitting can capture the most important and linear temperature change trend in the entire "Search" stage, and can better reflect the average heat release power in this stage than the average rate between two points. The physical meaning is clearer and more reliable. In addition, the data set identified by the algorithm analysis can also obtain a variety of indicators such as the longest continuous heating segment and the maximum temperature rise rate, to meet different research and testing needs;

[0093] (5) Reduced hardware dependence: Since this method (especially when combined with LOESS smoothing) has a high tolerance for high noise data, it can extract effective trends from signals with poor quality. This means that the signal-to-noise ratio requirements of ARC temperature acquisition systems (such as thermocouples) can be reduced to a certain extent, thereby potentially reducing equipment costs and maintenance difficulties, and improving the economy and universality of this analysis method.

[0094] The following will be combined with the appendix Figure 2 This application provides a detailed description of an ARC test temperature rise rate calculation device according to embodiments. It should be noted that... Figure 2 The illustrated ARC test temperature rise rate calculation device is used to perform the functions described in this application. Figure 1 The methods shown in the embodiments are illustrated for ease of explanation, showing only the parts relevant to the embodiments of this application. For specific technical details not disclosed, please refer to this application. Figure 1 The example shown.

[0095] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of an ARC test temperature rise rate calculation device provided in an embodiment of this application. Figure 2 As shown, the system includes a data acquisition module 201, a smoothing processing module 202, a robust regression module 203, and a slope acquisition module 204;

[0096] The data acquisition module 201 is used to acquire the raw time and temperature data during the ARC test search phase;

[0097] The smoothing module 202 is used to smooth the original time-temperature data based on a non-parametric smoothing method to obtain a central trend curve, and to scan the central trend curve based on a change point detection algorithm to extract multiple data segments, each of which includes a series of data points.

[0098] The robust regression module 203 is used to fit all data segments based on the robust regression algorithm to obtain the target linear model;

[0099] The slope acquisition module 204 is used to acquire the slope of the target linear model and use the slope as the temperature rise rate in the ARC test search phase.

[0100] In the real-time example of this application, the smoothing module 202 performs data smoothing segmentation. The original time-temperature data is the entire time-temperature data of the ARC test search phase. The smoothing module 202 smooths the entire time-temperature data of the ARC test search phase using a non-parametric smoothing method. A change point detection algorithm automatically identifies and extracts the data segments to be analyzed. Each data segment contains a series of (time t_i, temperature T_i) data points. The robust regression module 203 performs linear model fitting. For each data segment, a robust regression algorithm is applied to fit a target linear model T = a*t + b, where T is temperature and t is time. The slope acquisition module 204 extracts the slope and outputs the result. From the fitted target linear model T = a*t + b, the slope 'a' is extracted and used as the optimized temperature rise rate (dT / dt) for this "Search" phase, for subsequent thermal stability analysis. The smoothing module 202 smooths the original time-temperature data using a nonparametric smoothing method, addressing nonlinear dynamics and high-frequency noise issues. The robust regression module 203 fits all data segments using a robust regression algorithm, resolving interference from spurious signals and outliers, and achieving a successful fit. This overcomes interference from noise, outliers, and spurious signals, enabling more accurate and robust calculation of the temperature rise rate during the "Search" stage.

[0101] In one possible implementation, the nonparametric smoothing method is the LOESS algorithm; the smoothing module 202 performs smoothing processing on the original time-temperature data based on the nonparametric smoothing method to obtain a central trend curve, specifically including: setting a local window, wherein the number of data points in the local window accounts for a preset proportion of the total number of data points; sliding the local window along a preset direction to traverse the entire original time-temperature data, and performing weighted linear regression on multiple original data points located in the window each time using the LOESS algorithm to obtain multiple smoothed data points; and generating a smoothed central trend curve based on the multiple smoothed data points.

[0102] In this embodiment, the nonparametric smoothing method is the LOESS algorithm. The smoothing module 202 smooths the original time-temperature data using the nonparametric smoothing method to obtain a central trend curve, thus solving the problems of nonlinear dynamics and high-frequency noise. A preset ratio can be 10%, meaning the number of data points within a local window accounts for 10% of the total number of data points. The LOESS algorithm performs weighted linear regression on the original data points within this local window to obtain multiple smoothed data points corresponding to these original data points. Each time the local window is slid, the original data points within it are updated. The LOESS algorithm then performs weighted linear regression on these original data points to obtain multiple smoothed data points corresponding to these original data points. This process continues until the entire original time-temperature data is traversed by sliding the local window, and multiple smoothed data points corresponding to all original data points are obtained. Based on these multiple smoothed data points, a smooth central trend curve is generated. This central trend curve accurately reflects the macroscopic contour of the exothermic process and effectively filters out high-frequency random noise. The smoothing module 202 scans the central trend curve based on a change point detection algorithm to extract multiple data segments. Each data segment includes a series of data points. Specifically, it calls a change point detection algorithm (e.g., an algorithm based on abrupt changes in data slope and noise level) to scan the entire data sequence, automatically identify and segment all independent temperature change stages, and extract them into independent data segments D = {(t1, T1), (t2, T2), ..., (t_n, T_n)}.

[0103] In one possible implementation, the robust regression algorithm is the RANSAC algorithm; the robust regression module 203 includes a threshold setting unit, an iterative calculation unit, an interior point set unit, and a fitting unit;

[0104] The threshold setting unit is used to set the tolerance threshold for each data segment respectively;

[0105] The iterative calculation unit is used to perform iterative calculations on each data segment based on the tolerance threshold to obtain the data segment intra-point set corresponding to the data segment.

[0106] The inner point set unit is used to combine all the inner point sets of the data segments to form the optimal inner point set;

[0107] The fitting unit is used to fit the data points in the optimal interior point set based on the least squares method to obtain the target linear model.

[0108] In this embodiment, the threshold setting unit is used to calculate the absolute deviation of the median for each data segment, and multiply the absolute deviation of the median by a preset constant to obtain the tolerance threshold corresponding to the data segment.

[0109] In this embodiment, the iterative calculation unit is used to randomly extract a subset from the data segment, construct the current linear model based on the extracted subset, calculate the distance from all data points in the data segment to the current linear model, and mark data points whose distance is less than the tolerance threshold as the current inner point set; obtain the historical inner point set corresponding to the data segment, if there is no historical inner point set, then the current inner point set is used as the historical inner point set and the current linear model is used as the optimal model, if the number of points in the current inner point set is greater than the number of points in the historical inner point set, then the current inner point set is used as the historical inner point set and the current linear model is used as the optimal model; iteratively calculate until the number of iterations is a preset value, and then use the historical inner point set as the data segment inner point set corresponding to the data segment. Each iteration of the calculation process is as follows: a subset is randomly selected from the data segment, and the current linear model is constructed based on the subset. The distance from all data points in the data segment to the current linear model is calculated, and data points whose distance is less than the tolerance threshold are marked as the current inlier set. The historical inlier set corresponding to the data segment is obtained. If there is no historical inlier set, the current inlier set is used as the historical inlier set and the current linear model is used as the optimal model. If the number of points in the current inlier set is greater than the number of points in the historical inlier set, the current inlier set is used as the historical inlier set and the current linear model is used as the optimal model.

[0110] In this embodiment, the threshold setting unit is used for adaptive parameter setting: First, the median absolute deviation (MAD) of the data segment D is calculated. MAD = median (|T_i - median (T)|). Then, the tolerance threshold ε of RANSAC is automatically set based on the MAD value (e.g., ε = 2.5MAD). This allows the algorithm to adaptively determine whether a point is an interior point based on the noise level of the current data, and then use it for fitting. The median absolute deviation (MAD) of the data segment is calculated as MAD = median (|T_i - median (T)|). median: median, refers to the value in the middle position after arranging a set of data in ascending order. If the data volume is odd, the median value is taken; if it is even, the average of the two middle numbers is taken; |Ti - median (T)|: represents the absolute deviation (i.e., absolute distance) between each data point Ti and the median of the dataset T; median (|Ti - median (T)|) is the median of all absolute deviations, i.e., "median of absolute deviation". The RANSAC tolerance threshold ε is automatically set based on the median absolute deviation, where ε = k * MAD, k is a preset constant (e.g., 2.5), and MAD represents the median absolute deviation. This allows the algorithm to adaptively determine whether a point is an inlier based on the noise level of the current data. The median absolute deviation is a discrete metric insensitive to outliers; using it to set the threshold allows the algorithm to adapt to different data noise levels in different experiments and stages, thus achieving automated processing with minimal human intervention.

[0111] In this embodiment, the iterative computation unit is used to iteratively find the optimal model, and the algorithm enters an iterative loop. In each iteration, a minimum subset (2 points for a linear model) is randomly selected from the data to construct a temporary linear model M_temp. Then, the distance from all data points to this model is calculated, and points with a distance less than the tolerance threshold ε are marked as the current inlier set I_temp. The size of the current inlier set I_temp is compared with the size of the historical inlier set I_best, that is, the number of points in the current inlier set is compared with the number of points in the historical inlier set. If |I_temp|>|I_best|, then I_best=I_temp is updated. That is, if the number of points in the current inlier set is greater than the number of points in the historical inlier set, then the current inlier set is used as the historical inlier set, and the current linear model M_temp is recorded as the optimal model M_best. This iterative process is repeated a specified number of times, such as 100 times. 100 times is a robust choice, and the process of finding the optimal fitting parameters can be completed faster. The preset value can be 100. When the number of iterations is a preset value, the historical inner point set is used as the inner point set corresponding to the data segment. For each data segment, the inner point set corresponding to the data segment is obtained through the iterative calculation unit. The inner point set unit combines the inner point sets corresponding to all data segments to form the optimal inner point set.

[0112] In this embodiment, the fitting unit is used to perform the final model fitting. Using all data points in the optimal interior set I_best, a final and most robust linear model T=at+b is fitted using the standard least squares method.

[0113] In this embodiment, the data acquisition module 201 is used to load a complete ARC test data file through a data interface; and to filter the raw time and temperature data of the ARC test search stage from the ARC test data file according to the stage identifier. The complete ARC test data file includes data corresponding to each stage of the ARC test, and the raw time and temperature data of the ARC test search stage is filtered from the ARC test data file according to the stage identifier. A complete ARC test data file is loaded through the data interface, and the file contains a sequence of timestamps (t) and corresponding temperature values ​​(T) sampled at fixed time intervals.

[0114] Those skilled in the art will clearly understand that the technical solutions of the embodiments of this application can be implemented by means of software and / or hardware. In this specification, "unit", "module" and "part" refer to software and / or hardware that can independently complete or cooperate with other components to complete a specific function, wherein the hardware may be, for example, a field-programmable gate array (FPGA), an integrated circuit (IC), etc.

[0115] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0116] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0117] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between devices or units may be electrical or other forms.

[0118] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0119] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0120] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0121] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0122] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of embodiments of this disclosure upon considering the specification and practicing the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described herein. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for calculating the temperature rise rate in ARC testing, characterized in that, Includes the following steps: S1: Obtain the raw time and temperature data for the ARC test search phase; S2: The original time-temperature data is smoothed using a non-parametric smoothing method to obtain a central trend curve. The central trend curve is then scanned using a change point detection algorithm to extract multiple data segments, each of which includes a series of data points. S3: Fit all data segments using a robust regression algorithm to obtain the target linear model; S4: Obtain the slope of the target linear model and use this slope as the temperature rise rate in the ARC test search phase.

2. The method for calculating the temperature rise rate in an ARC test as described in claim 1, characterized in that, The nonparametric smoothing method in step S2 is the LOESS algorithm, and the robust regression algorithm in step S3 is the RANSAC algorithm.

3. The method for calculating the temperature rise rate in an ARC test as described in claim 2, characterized in that, The process of smoothing raw time-temperature data using the LOESS algorithm to obtain a central trend curve specifically includes: setting a local window, wherein the number of data points within the local window is a preset proportion of the total number of data points; sliding the local window along a preset direction to traverse the entire raw time-temperature data; performing weighted linear regression on multiple raw data points within the window each time using the LOESS algorithm to obtain multiple smoothed data points; and generating a smooth central trend curve based on the multiple smoothed data points.

4. The method for calculating the ARC test temperature rise rate as described in claim 2 or 3, characterized in that, Step S3 specifically includes: S31: Set the tolerance threshold for each data segment; S32: Based on the tolerance threshold, iteratively calculate each data segment to obtain the point set within the corresponding data segment; S33: Combine all the point sets within the data segments to form the optimal internal point set; S34: Fit the data points in the optimal interior point set using the least squares method to obtain the target linear model.

5. The method for calculating the ARC test temperature rise rate as described in claim 4, characterized in that, Step S31 specifically includes: S311: Calculate the absolute deviation of the median for each data segment; S312: Calculate the tolerance threshold corresponding to the data segment by multiplying the median absolute deviation by a preset constant.

6. The method for calculating the temperature rise rate in an ARC test as described in claim 4, characterized in that, Step S32 specifically includes: S321: Randomly select a subset from the data segment and construct the current linear model based on the selected subset; S322: Calculate the distance from all data points in the acquired data segment to the current linear model, and mark the data points whose distance is less than the tolerance threshold as the current inner point set; S323: Obtain the historical inner point set corresponding to the data segment. If there is no historical inner point set, use the current inner point set as the historical inner point set and the current linear model as the optimal model. If the number of points in the current inner point set is greater than the number of points in the historical inner point set, use the current inner point set as the historical inner point set and the current linear model as the optimal model. S324: Return to S321 for iterative calculation until the number of iterations is the preset value, and use the historical point set as the point set of the corresponding data segment.

7. A method for calculating the temperature rise rate in an ARC test as described in claim 1, 2, or 3, characterized in that: Step S1 specifically includes: loading an ARC test data file; and filtering the raw time and temperature data of the ARC test search phase from the ARC test data file according to the phase identifier.

8. An ARC test temperature rise rate calculation device, characterized in that: It includes a data acquisition module, a smoothing module, a robust regression module, and a slope acquisition module; The data acquisition module is used to acquire the raw time and temperature data during the ARC test search phase; The smoothing module is used to smooth the original time-temperature data based on a non-parametric smoothing method to obtain a central trend curve, and to scan the central trend curve based on a change point detection algorithm to extract multiple data segments, each of which includes a series of data points. The robust regression module is used to fit all data segments based on the robust regression algorithm to obtain the target linear model; The slope acquisition module is used to obtain the slope of the target linear model and use the slope as the temperature rise rate in the ARC test search phase.

9. The ARC test temperature rise rate calculation device as described in claim 8, characterized in that, The nonparametric smoothing method is the LOESS algorithm. The smoothing module uses the nonparametric smoothing method to smooth the original time-temperature data to obtain a central trend curve. Specifically, this includes: setting a local window, where the number of data points within the local window is a preset proportion of the total number of data points; sliding the local window along a preset direction to traverse the entire original time-temperature data; using the LOESS algorithm to perform weighted linear regression on multiple original data points within the window each time to obtain multiple smoothed data points; and generating a smoothed central trend curve based on the multiple smoothed data points.

10. An ARC test temperature rise rate calculation device as described in claim 8 or 9, characterized in that, The robust regression algorithm is the RANSAC algorithm; the robust regression module includes a threshold setting unit, an iterative calculation unit, an interior point set unit, and a fitting unit; The threshold setting unit is used to set the tolerance threshold for each data segment respectively; The iterative calculation unit is used to perform iterative calculations on each data segment based on the tolerance threshold to obtain the data segment intra-point set corresponding to the data segment. The inner point set unit is used to combine all the inner point sets of the data segments to form the optimal inner point set; The fitting unit is used to fit the data points in the optimal interior point set based on the least squares method to obtain the target linear model.

Images