A data processing method for a thermal treatment process

Abstract

The present application relates to the technical field of data processing, and more particularly, to a data processing method for a heat treatment process, comprising: obtaining each process data of the heat treatment process respectively, and obtaining final components of each process data by independent component analysis to realize data processing for the heat treatment process; in each iteration process of the independent component analysis, in response to a change feature of all components being less than a preset threshold, iteration is terminated, and the components at this time are taken as final components, comprising: calculating a stable feature of any data point in any component, a stationarity of any component, a curve similarity of two components, a separation degree of two components and a change feature of all components respectively. The present application dynamically judges whether to stop iteration according to the change feature of the components, avoids early stopping or over-iteration caused by fixed iteration times, and saves computing resources.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing. More specifically, the present application relates to a data processing method for heat treatment process. BACKGROUND

[0002] Heat treatment process involves multiple parameters and steps, such as heating, holding, cooling, etc., and the selection of parameters for each step directly affects the performance and quality of the final product. Due to the complexity of heat treatment process, traditional data processing methods often fail to comprehensively and accurately capture and analyze key information in the process. During the heat treatment process, sensors collect a large amount of data, which often contains redundant information and noise, interfering with the accuracy and reliability of the data.

[0003] Independent Component Analysis (ICA) is an algorithm widely used in signal processing, image processing and other fields, which focuses on recovering independent components from observed multivariate signals; this feature makes ICA have a significant advantage in processing mixed signals, and can separate independent components directly related to the heat treatment process, thereby removing redundancy and noise.

[0004] Patent document No. CN118094210B discloses a method for identifying the charging and discharging behavior of an energy storage system based on underdetermined blind source separation, which relates to the technical field of energy storage charging and discharging behavior identification, and solves the problem of low perception level of energy storage demand response capability in the prior art. The method includes: obtaining a total power signal sequence of the energy storage system, using an ensemble empirical mode decomposition algorithm to perform signal dimensionality increase on the total power signal sequence to obtain an intrinsic mode function matrix; using a principal component analysis algorithm to analyze the intrinsic mode function matrix to estimate the number of independent energy storage signals and obtain a non-underdetermined dimensionality increased signal matrix corresponding to the total power signal sequence; analyzing the independence between each non-underdetermined dimensionality increased signal in the non-underdetermined dimensionality increased signal matrix to obtain a corresponding independent sub-signal group matrix; and performing blind source signal separation on the independent sub-signal group matrix based on a fast independent component analysis method to obtain the charging and discharging behavior curves of each independent energy storage in the energy storage system.

[0005] However, the above patent document does not solve the problem that the independent component analysis method needs to preset the number of iterations during use, and the complexity of the heat treatment process and the variability of the environment result in complex heat treatment data, so the preset number of iterations cannot be adjusted according to the complexity of the data and the convergence speed, which may cause the algorithm to stop too early when it has not reached the best performance, or continue to iterate after it has converged, causing resource waste. SUMMARY

[0006] To address the issue that independent component analysis (ICA) requires a preset number of iterations, and the complexity of heat treatment processes and the variability of the environment lead to complex heat treatment data, the preset number of iterations cannot be adjusted according to the complexity of the data and the convergence speed. This may cause the algorithm to stop prematurely before reaching optimal performance, or continue iterating after convergence, resulting in wasted resources. Therefore, this invention proposes a data processing method for heat treatment processes, which includes the following steps:

[0007] Each process data point of the heat treatment process is acquired separately, and the final components of each process data point are obtained using independent component analysis (ICA) to achieve data processing for the heat treatment process. In each iteration of ICA, if the variation characteristics of all components are less than a preset threshold, the components of that process data point are considered stable, the iteration terminates, and the components at this point are taken as the final components. Otherwise, the next iteration begins. This includes: designating any component separated by ICA as the target component; calculating the stability characteristics of any data point in the target component based on the Euclidean distance between any data point and its two adjacent data points; performing clustering based on the stability characteristics of each data point in the target component, and calculating the stationarity of the target component based on the number and size of the clusters; performing curve fitting on the data points of all components to obtain a fitted curve for each component, and calculating the curve similarity between two components based on the area enclosed by the fitted curves of any two components; calculating the separation degree between two components based on the curve similarity and DTW similarity; and calculating the variation characteristics of all components based on the separation degree and stationarity.

[0008] By dynamically determining whether to stop iteration based on the changing characteristics of components, premature stopping or excessive iteration caused by a fixed number of iterations is avoided, thus saving computational resources. By evaluating the stability of multiple dimensions and calculating the degree of component separation, the stability of the data is accurately determined, ensuring efficient convergence of the algorithm, reducing unnecessary calculations, improving algorithm efficiency, and reducing energy consumption and computational resource consumption.

[0009] Furthermore, the stable feature satisfies the following relationship:

[0010] ;

[0011] In the formula, For the first The first round of iteration The first of the components Stable characteristics of each data point For the first The first round of iteration The first of the components a Euclidean distance between the data point and its two adjacent data points, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration.

[0012] Further, the cluster shape size satisfies the following relationship:

[0013] ;

[0014] In the formula, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration.

[0015] Further, the stationarity satisfies the following relationship:

[0016] ;

[0017] In the formula, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, is the number of the iteration, The number of clusters in the clustering results of each component. It is a natural exponential function.

[0018] By introducing first-order differences in cluster shape and size and processing them using the natural exponential function, the stationarity of each component during the iteration process can be measured more accurately. This effectively captures subtle changes in the data and improves the accuracy of stationarity assessment.

[0019] Furthermore, the curve fitting is performed using the least squares method.

[0020] Furthermore, the curve similarity satisfies the following relationship:

[0021] ;

[0022] In the formula, For the first The first round of iteration The first component and the first Curve similarity of components For the first The first round of iteration The first component and the first The area of ​​the figure enclosed by the fitted curves of each component. For the first The first round of iteration Fitting curves for each component, For the first The first round of iteration Fitting curves for each component, This refers to the initial data acquisition time for the process data of the component category. This refers to the last time the process data for the component category was collected. It is the absolute value symbol.

[0023] By calculating the area difference between fitted curves, the similarity between different components can be measured more accurately, avoiding the limitations of simple numerical comparison and effectively capturing subtle differences between components.

[0024] Furthermore, the degree of separation satisfies the following relationship:

[0025] ;

[0026] In the formula, For the first The first round of iteration The first component and the first The degree of separation of the components For the first The first round of iteration The first component and the first Curve similarity of components For the first The first round of iteration The first component and the first DTW similarity of each component.

[0027] By quantifying the degree of separation, the differences between different components can be measured more accurately; by combining curve similarity and DTW similarity, the similarity between components can be effectively distinguished, thereby effectively dealing with noise and fluctuations that may exist in the data and improving the stability and robustness of the algorithm when processing complex data.

[0028] Furthermore, the changing characteristics satisfy the following relationship:

[0029] ;

[0030] In the formula, for The characteristics of changes in all components during the round of iterations, For the first The first round of iteration The first component and the first The degree of separation of the components For the first The first round of iteration The stability of each component For the first The first round of iteration The stability of each component For the first The number of all components during the round of iteration, It is a natural exponential function.

[0031] By combining the degree of separation and component stability, the interaction and separation effect between each pair of components can be accurately measured. In multiple iterations, the change characteristics of the components are gradually optimized, making the final separation effect more accurate and applicable to complex multi-component systems. By introducing the stability of the components as a regulating factor, deviations caused by instability or noise in the system during the separation process can be avoided.

[0032] Furthermore, the data processing for the heat treatment process includes: taking the mean value of the DTW similarity between the target component and the other components as the similarity of the target component, recording the component corresponding to the minimum value in the similarity as a noise component and removing it, thus completing the data processing for the heat treatment process.

[0033] The present invention has the following beneficial effects:

[0034] Since in the independent component analysis method, the preset iteration number cannot be adjusted according to the complexity of data and the convergence speed, the algorithm is stopped too early when the optimal performance is not reached, or the iteration is continued after convergence, causing resource waste, by setting appropriate convergence conditions, the algorithm can be stopped in time at the optimal performance, avoid inefficient iteration caused by the preset number, ensure the best result, thereby improving the denoising effect; meanwhile, the present application can cope with the complex data and variable environment in the heat treatment process, and improves the accuracy and reliability of data processing. BRIEF DESCRIPTION OF DRAWINGS

[0035] The above and other objects, features and advantages of the present application will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings in which a number of embodiments of the application are shown by way of illustration, now described in detail. In the drawings, identical or similar components are denoted with the same reference numerals, and:

[0036] Figure 1 is a step flow chart of a data processing method for a heat treatment process according to an embodiment of the present application.

[0037] Figure 2 is a step flow chart of a data processing method for a heat treatment process in each iteration process of independent component analysis according to an embodiment of the present application. DETAILED DESCRIPTION

[0038] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of, rather than all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0039] The specific embodiments of the present application will be described in detail below with reference to the drawings.

[0040] Please refer to Figure 1 which shows a step flow chart of a data processing method for a heat treatment process according to an embodiment of the present application, the method comprising the following steps:

[0041] S1: respectively acquiring each process data of the heat treatment process.

[0042] By installing various types of sensors to collect data in the heat treatment process, such as temperature, time, toughness, etc., different collection frequencies are set for different types of processed data, for example, temperature data is analyzed, and the implementation personnel can set the data collection frequency according to the specific implementation conditions, for example, 1Hz.

[0043] S2: obtaining final components of each process data by independent component analysis method.

[0044] It should be noted that due to the complexity of the heat treatment process and the variability of the environment, the heat treatment data is relatively complex, and when the independent component analysis separates each component through iteration, the preset number of iterations cannot be adjusted according to the complexity of the data and the convergence speed, which may cause the algorithm to stop too early when the optimal performance is not reached, or continue to iterate after convergence, thereby causing the final component separation effect to be poor, affecting the removal of noise components, and resulting in unsatisfactory heat treatment data denoising effect. The change characteristics of the obtained components can directly reflect the convergence of the algorithm, and when the algorithm gradually converges to the optimal solution, the change amount of the obtained components will tend to be stable or gradually decrease. By monitoring the change characteristics of the obtained components, the iteration process can be accurately controlled; when it is found that the change amount of the obtained components reaches the preset threshold, the iteration can be terminated in time, thereby ensuring the accuracy and reliability of the results, and helping to avoid errors and uncertainties caused by insufficient or excessive iterations. Because the independent component analysis method processes one-dimensional data by dividing the data into multiple independent components, as the iteration proceeds, the multiple components will gradually stabilize, which means that the algorithm has converged, and further iteration will not bring significant changes. At the same time, as the iteration proceeds, the separation degree between the multiple components will gradually increase, which means that the model has effectively extracted the differences between various features, and further iteration may not have further improvement. Therefore, by analyzing the change characteristics of the components in the iteration process, the stability of the components and the change characteristics between the components can be considered. In summary, the present application analyzes the change characteristics of the components from the stability of the components in the iteration process and the change characteristics between the components, and uses them as the basis for terminating the iteration, and then obtains the noise components to realize the denoising of the heat treatment data.

[0045] In each iteration process of the independent component analysis method, please refer to Figure 2 , in response to the change characteristics of all components being less than the preset threshold, it is determined that the components of the process data are stable, the iteration is terminated, and the components at this time are taken as the final components, otherwise, the next iteration is entered, including:

[0046] The threshold can be set by the implementer according to the specific implementation, for example, 0.5.

[0047] S201: calculating the stability feature of any data point in any component.

[0048] Any component separated by the independent component analysis method is referred to as a target component; based on the Euclidean distance between any data point in the target component and the two data points adjacent to the data point, the stability feature of any data point in the target component is calculated.

[0049] Specifically, the stability feature satisfies the following relationship:

[0050] ;

[0051] In the formula, is the stability feature of the i th data point in the j th component in the k th iteration process, is the Euclidean distance between the i th data point and its adjacent two data points in the j th component in the k th iteration process, is the value of the i th data point in the j th component in the k th iteration process, is the value of the i th data point in the j th component in the k th iteration process, is the value of the i th data point in the j th component in the k th iteration process.

[0052] wherein, represents the Euclidean distance between the i th data point and its adjacent two data points in the j th component in the k th iteration process, the smaller the distance is, the higher the stability feature of the data is. S202: Calculate the stationarity of any one component. It should be noted that when obtaining the stationarity of each component, the clustering result of the data can be used to reflect it, but in many practical applications, the original data is often affected by factors such as sensor error, resulting in noise in the data, which will interfere with the effect of the clustering algorithm, making the clustering result inaccurate. The stationarity feature is usually calculated based on the change trend between data, the distance between adjacent data points, etc., and can reflect the inherent stability of the data. By clustering the stationarity feature, it can help to remove unstable data points and noise, and focus on the actual trend and stability of the data, so that the effect of clustering will be more accurate and reliable.

[0053] According to the stability feature of each data point in the target component, clustering is performed, and based on the number and shape size of the clusters in the clustering result, the stationarity of the target component is calculated.

[0054]

[0055]

[0056] ​​​​​​​​​​​​​​​​​​Specifically, the shape and size of the clusters satisfy the following relationship:

[0057] ;

[0058] In the formula, For the first The first round of iteration The first component in the clustering results The shape and size of each cluster For the first The first round of iteration The first component in the clustering results The maximum value of data points in each cluster. For the first The first round of iteration The first component in the clustering results The minimum value of data points in each cluster. For the first The first round of iteration The first component in the clustering results The number of data points in each cluster.

[0059] in, Indicates the first The first round of iteration The first component in the clustering results Each cluster corresponds to a data range. The smaller the data range and the more data in the cluster, the denser the data in the cluster and the smaller the shape of the cluster. No constraints are imposed on the clustering algorithm during the clustering process; the clustering result is sufficient. For example, the DBSCAN clustering algorithm.

[0060] Specifically, the stationarity satisfies the following relationship:

[0061] ;

[0062] In the formula, For the first The first round of iteration The stability of each component For the first The first iteration in the round The first component in the clustering results First-order difference values ​​of the shape and size of each cluster For the first The first round of iteration The number of clusters in the clustering results of each component. It is a natural exponential function.

[0063] in, The first The first The mean value of the first-order difference of the shape and size of the cluster in the data stable feature clustering result of the component, the smaller the value, the higher the consistency of the data stable feature in the component, and the higher the stationarity of the component.

[0064] S203: Calculate the curve similarity of the two components.

[0065] It should be noted that when obtaining the change feature between any two components, the correlation of the corresponding data of the two components and the similarity of the corresponding curves are combined to more comprehensively evaluate the change feature between the two components, which can avoid the one-sidedness that may be caused by a single indicator. For example, even if the similarity of the two components is low in data, if there is a strong correlation between them, this difference is not enough to indicate that the change feature between them is high.

[0066] The data points in all components are curve fitted to obtain the fitting curve of each component, and the curve similarity of the two components is calculated based on the area of the figure enclosed by the fitting curves of any two components.

[0067] Specifically, the curve fitting is curve fitting by the least square method.

[0068] Specifically, the curve similarity satisfies the following relationship:

[0069] ;

[0070] In the formula, The curve similarity of the first The first The first The curve similarity of the first The area of the figure enclosed by the fitting curves of the first The first The first The area of the figure enclosed by the fitting curves of the first The fitting curve of the first The fitting curve of the first The fitting curve of the first The fitting curve of the first The fitting curve of the first The initial collection time of the process data to which the component belongs, The last collection time of the process data to which the component belongs, The absolute value symbol.

[0071] Wherein, The first The first​ the area of the figure enclosed by the fitting curve of the first component and the fitting curve of the second component, the smaller the area is, the higher the similarity of the two curves is.

[0072] S204: Calculate the separation degree of the two components.

[0073] It should be noted that, in obtaining the similarity of the two component corresponding data, and taking the similarity of the two component corresponding data and the similarity of the corresponding curves as the index for calculating the separation degree between the two components, the lower the similarity of the two component corresponding data is and the smaller the similarity of the corresponding curves is, the higher the separation degree of the two components is.

[0074] Based on the curve similarity and the DTW similarity of any two components, the separation degree of the two components is calculated.

[0075] Specifically, the separation degree satisfies the following relationship:

[0076] ;

[0077] In the formula, is the separation degree of the first component and the second component in the first iteration process, is the separation degree of the first component and the second component in the first iteration process, is the curve similarity of the first component and the second component in the first iteration process, is the curve similarity of the first component and the second component in the first iteration process, is the DTW similarity of the first component and the second component in the first iteration process, is the DTW similarity of the first component and the second component in the first iteration process.

[0078] S205: Calculate the change feature of all components.

[0079] It should be noted that, the higher the stability of the two components is and the higher the separation degree between the two components is, the higher the stability of each component in the current iteration process is, and the smaller the change feature of the components obtained in the iteration process is.

[0080] Based on the separation degree and the stability, the change feature of all components is calculated.

[0081] Specifically, the change feature satisfies the following relationship:

[0082] ;

[0083] In the formula, is the change feature of all components in the first iteration process, is the change feature of all components in the first iteration process, is the change feature of all components in the first iteration process.​​​​​​​ The first round of iteration The first component and the first The degree of separation of the components For the first The first round of iteration The stability of each component For the first The first round of iteration The stability of each component For the first The number of all components during the round of iteration, It is a natural exponential function.

[0084] S3: To enable data processing for heat treatment processes.

[0085] Specifically, the data processing for the heat treatment process includes:

[0086] The mean value of the DTW similarity between the target component and the other components is taken as the similarity of the target component. The component corresponding to the minimum value of the similarity is recorded as a noise component and removed, thus completing the data processing for the heat treatment process.

[0087] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A data processing method for heat treatment processes, characterized in that, include: Each process data of the heat treatment process is obtained separately, and the final components of each process data are obtained by independent component analysis to realize data processing for heat treatment process. In each iteration of the independent component analysis method, if the change characteristics of all components are less than a preset threshold, the components of the process data are considered stable, the iteration terminates, and the components at this point are taken as the final components. Otherwise, the next iteration begins, including: Any component separated by independent component analysis is denoted as the target component; the stability feature of any data point in the target component is calculated based on the Euclidean distance between any data point in the target component and its two adjacent data points; clustering is performed based on the stability feature of each data point in the target component, and the stationarity of the target component is calculated based on the number of clusters and the size of the clusters in the clustering results. Curve fitting is performed on the data points of all components to obtain the fitted curve for each component. The curve similarity between two components is calculated based on the area enclosed by the fitted curves of any two components. The separation degree between two components is calculated based on the curve similarity and DTW similarity. The variation characteristics of all components are calculated based on the separation degree and stationarity. The changing characteristics satisfy the following relationship: ； In the formula, for The characteristics of changes in all components during the round of iterations, For the first The first round of iteration The first component and the first The degree of separation of the components For the first The first round of iteration The stability of each component For the first The first round of iteration The stability of each component For the first The number of all components during the round of iteration, It is a natural exponential function.

2. The data processing method for heat treatment process according to claim 1, characterized in that, The stable feature satisfies the following relationship: ; In the formula, For the first The first round of iteration The first of the components Stable characteristics of each data point For the first The first round of iteration The first of the components The Euclidean distance between a data point and its two adjacent data points For the first The first round of iteration The first of the components The value of each data point. For the first The first round of iteration The first of the components The value of each data point. For the first The first round of iteration The first of the components The value of each data point.

3. The data processing method for heat treatment process according to claim 1, characterized in that, The shape and size of the clusters satisfy the following relationship: ； In the formula, For the first The first round of iteration The first component in the clustering results The shape and size of each cluster For the first The first round of iteration The first component in the clustering results The maximum value of data points in each cluster. For the first The first round of iteration The first component in the clustering results The minimum value of data points in each cluster. For the first The first round of iteration The first component in the clustering results The number of data points in each cluster.

4. The data processing method for heat treatment process according to claim 1, characterized in that, The stationarity satisfies the following relationship: ； In the formula, For the first The first round of iteration The stability of each component For the first The first round of iteration The first component in the clustering results First-order difference values ​​of the shape and size of each cluster For the first The first round of iteration The number of clusters in the clustering results of each component. It is a natural exponential function.

5. The data processing method for heat treatment process according to claim 1, characterized in that, The curve fitting is performed using the least squares method.

6. The data processing method for heat treatment process according to claim 1, characterized in that, The curve similarity satisfies the following relationship: ； In the formula, For the first The first round of iteration The first component and the first Curve similarity of components For the first The first round of iteration The first component and the first The area of ​​the figure enclosed by the fitted curves of each component. For the first The first round of iteration Fitting curves for each component, For the first The first round of iteration Fitting curves for each component, This refers to the initial data acquisition time for the process data of the component category. This refers to the last time the process data for the component category was collected. It is the absolute value symbol.

7. The data processing method for heat treatment process according to claim 1, characterized in that, The degree of separation satisfies the following relationship: ; In the formula, For the first The first round of iteration The first component and the first The degree of separation of the components For the first The first round of iteration The first component and the first Curve similarity of components For the first The first round of iteration The first component and the first DTW similarity of each component.

8. A data processing method for heat treatment processes according to claim 1, characterized in that, The data processing for the heat treatment process includes: The mean value of the DTW similarity between the target component and the other components is taken as the similarity of the target component. The component corresponding to the minimum value of the similarity is recorded as a noise component and removed, thus completing the data processing for the heat treatment process.

Images