Cooperative heating control method for infrared hot box of false twist texturing machine

Abstract

The invention discloses a cooperative heating control method and system for an infrared hot box of a false twist texturing machine, and relates to the field of chemical fiber textile machinery. Synchronously acquiring cross section temperature distribution data and material state spectral data of the tows, and respectively performing spatial domain and frequency domain analysis to extract characteristic parameters; fusing and generating a composite state index based on the extracted feature parameters; by utilizing the index, combining heat transfer model prediction and expert rule base reasoning, and through an arbitration algorithm, generating a collaborative power adjustment instruction for each independent heating partition; independently adjusting partition power according to the instruction; and finally optimizing control parameters based on historical production data. According to the invention, intelligent decision and closed-loop control of multi-source information fusion are realized, the problems of non-uniform heating of the cross section of the tow, control response lag and isolated setting of process parameters are effectively solved, and the heating efficiency, the control precision and the consistency of product quality are improved.

Description

Technical Field

[0001] This invention relates to the field of chemical fiber textile machinery technology, and in particular to a method for coordinated heating control of infrared heating boxes in a false twist texturing machine. Background Technology

[0002] The false-twist texturing machine is a key piece of equipment in chemical fiber production. It involves heating and twisting the yarn, followed by cooling and untwisting. The resulting yarn is called "textured elastic yarn." The heating box is the core component of the texturing machine. Its function is to precisely and uniformly heat the yarn bundle during operation, plasticizing it and reducing tensile stress. Its performance directly affects the quality of the final finished yarn (such as crimp properties, elasticity, and dyeing uniformity). Currently, existing heating boxes generally use electric heating tubes for contact heating or high-temperature heat medium heating. This technology has the following shortcomings: 1. Limited information perception: The use of single or limited temperature measurement points makes it impossible to obtain the complete temperature distribution of the cross-section of the fiber bundle, and even more impossible to directly and online perceive the changes in the microscopic physical state of the fiber bundle material itself caused by changes in molecular structure during the heating process (such as changes in crystallinity and orientation).

[0003] 2. Simple decision-making logic: It mostly adopts PID control based on the deviation between the fixed set point and the measured point, which is a "post-event remedial" feedback. It cannot predict the heating trend and is slow to adjust when faced with changes in wire speed and environmental interference, which can easily cause temperature fluctuations.

[0004] 3. Lack of system coordination: The heating parameters of the hot box and the process parameters of the false twister (such as the D / Y ratio) are usually set manually. There is a lack of automated correlation and coordination adjustment mechanism between the two, making it difficult to achieve the best global processing effect.

[0005] Therefore, there is an urgent need for a new heating control method that can integrate multi-dimensional information, possess intelligent decision-making and prediction capabilities, and achieve cross-subsystem process collaboration. Summary of the Invention

[0006] To address these issues, this invention provides a method for coordinated heating control of an infrared heating box in a false twisting deformation machine. This method aims to control problems such as one-sided state judgment due to a single source of information, lagging control strategies and lack of intelligent decision-making capabilities, difficulty in coordinating the isolated setting of heating process and subsequent processing parameters, and insufficient adaptive optimization capabilities of the system.

[0007] To achieve the above objectives, the present invention provides a method for coordinated heating control of an infrared heating box in a false-twist texturing machine, comprising the following steps: Step S1: Set the initial power of each heating zone according to the product specifications of the filament bundle to be processed; Step S2: Collect cross-sectional temperature distribution data of the filament bundle passing through the hot box, and perform spatial domain analysis on the temperature distribution data to extract temperature uniformity characteristic parameters. Step S3: Collect the material state spectral data through the hot box, and perform frequency domain or characteristic wavelength analysis on the material state spectral data to extract material state characteristic data; Step S4: Determine the composite state index based on the temperature uniformity characteristic parameters and the material state characteristic data; Step S5: Based on the composite state index, heat transfer model, and rule base, generate a coordinated power adjustment command for at least two independent heating zones within the hot box; Step S6: Adjust the heating power of each heating zone independently according to the coordinated power adjustment command; Step S7: Optimize control parameters based on historical production data.

[0008] Furthermore, in step S1, the initial heating parameters corresponding to the current product specification are retrieved from the pre-stored set of process parameters to set the initial power of each heating zone; wherein, the set of process parameters is associated with and stores at least different product specifications and their corresponding initial heating parameters.

[0009] Furthermore, the data acquisition in steps S2 and S3 is performed by a synchronously triggered infrared thermal imaging device and an online spectrometer to ensure the temporal consistency between the temperature distribution data and the material state spectral data; the at least two independent heating zones include a central main heating zone and at least two edge compensation heating zones.

[0010] Further, in step S2, the spatial domain analysis of the temperature distribution data includes: calculating the standard deviation, coefficient of variation, or range of the temperature distribution in the cross section as the temperature uniformity characteristic parameter; and / or calculating the average temperature difference between a preset central region and an edge region in the cross section as the temperature uniformity characteristic parameter.

[0011] Further, in step S3, the frequency domain or characteristic wavelength analysis of the material state spectral data includes: identifying characteristic absorption peaks corresponding to the chemical bonds of the filament material, and calculating the intensity, peak area, or intensity ratio of the characteristic absorption peaks relative to a reference peak as the material state characteristic data; and / or calculating the derivative, slope, or curvature of the material state spectral data in a specific band as the material state characteristic data.

[0012] Further, in the step S4, the determining the composite state indicator comprises: performing normalization processing on the temperature uniformity characteristic parameter and the material state characteristic data respectively, and performing weighted fusion according to a preset weight coefficient to generate the composite state indicator in a scalar form.

[0013] Further, in the step S5, the generating the cooperative power adjustment instruction comprises the following sub-steps: Step S51, inputting the composite state indicator into the heat transfer model to predict future change trends of the yarn bundle state under different power adjustment schemes; Step S52, querying the rule base according to the composite state indicator to obtain a power adjustment strategy based on expert experience; Step S53, generating a final cooperative power adjustment instruction by comprehensively adopting a preset arbitration algorithm and the future change trends and the power adjustment strategy.

[0014] Further, the arbitration algorithm is configured to dynamically adjust weight proportions of the prediction result of the heat transfer model and the reasoning result of the rule base in the final instruction according to a confidence level of the composite state indicator or a change rate of the material state characteristic data.

[0015] Further, the set of process parameters further stores a false twister process parameter suggestion value linked with the initial heating parameter.

[0016] Further, in the step S7, the process of optimizing the control parameter comprises: Step S71, recording the composite state indicator, the executed cooperative power adjustment instruction and corresponding product quality data in historical production data; Step S72, analyzing the historical production data by a machine learning algorithm to optimize parameters of the heat transfer model, rules in the rule base or the weight coefficient in the step S4.

[0017] Compared with the prior art, the present application has the following beneficial effects: 1. Improved perception dimension and accuracy: by synchronously collecting infrared thermal imaging and online spectral analysis, fusing temperature field uniformity information and material crystallinity / orientation information, comprehensive and accurate judgment of the yarn processing state is realized.

[0018] 2. Intelligent forward-looking control is realized: a composite decision mechanism combining "heat transfer model prediction" and "expert rule base reasoning" is adopted, and a dynamic arbitration algorithm is introduced, so that the power adjustment has scientificity and experience, and the response is faster and more accurate.

[0019] 3. Promote process synergy optimization: by pre-storing the process parameter set associated with the false twister parameter, realize the intelligent matching and one-key calling of the heating parameter and the core process parameter, improve the synergy and stability of the whole machine process.

[0020] 4. With self-learning evolution ability: by recording production data and using machine learning for optimization, the system can continuously improve itself, adapt to different raw materials, specifications and product requirements, and reduce the dependence on operator experience. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 The step flow chart of the false twist texturing machine infrared heat box cooperative heating control method of the embodiment of the present application is shown in the figure. Figure 2 The step flow chart of generating a cooperative power adjustment instruction of the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0022] In order to make the purpose and advantages of the present application clearer and more apparent, the present application will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0023] The preferred embodiments of the present application will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present application and are not intended to limit the protection scope of the present application.

[0024] Please refer to Figure 1 shown, Figure 1 The step flow chart of the false twist texturing machine infrared heat box cooperative heating control method of the embodiment of the present application is shown in the figure.

[0025] The false twist texturing machine infrared heat box cooperative heating control method of the embodiment of the present application comprises the following steps: Step S1, according to the product specification of the to-be-processed yarn, set the initial power of each heating partition; Step S2, collect the cross-sectional temperature distribution data of the yarn passing through the heat box, and perform spatial domain analysis on the temperature distribution data to extract the temperature uniformity characteristic parameters; Step S3, collect the material state spectrum data of the material passing through the heat box, and perform frequency domain or characteristic wavelength analysis on the material state spectrum data to extract the material state characteristic data; Step S4, determine the composite state index based on the temperature uniformity characteristic parameters and the material state characteristic data; Step S5, based on the composite state index, the heat transfer model and the rule base, generate a cooperative power adjustment instruction for at least two independent heating partitions in the heat box; Step S6, independently adjust the heating power of each heating partition according to the cooperative power adjustment instruction; Step S7, based on historical production data, optimize control parameters.

[0026] Specifically, in step S1, from the pre-stored process parameter set, the initial heating parameters corresponding to the current product specification are called to set the initial power of each heating partition; wherein the process parameter set is at least associatedly stored with different product specifications and their corresponding initial heating parameters.

[0027] Specifically, the pre-stored process parameter set is stored in the form of a database in the collaborative control unit. For example, for a yarn with a specification of "150D / 48F polyester FDY", its corresponding record can include: the initial power of the center main heating zone , the initial power of the edge compensation heating zone , and the associated false twister D / Y ratio recommended range is [1.68, 1.72]. After the operator selects the product on the control interface, the system automatically calls these parameters and issues them.

[0028] Specifically, the data collection of steps S2 and S3 is performed by a synchronously triggered infrared thermal imaging device and an online spectrometer to ensure the time consistency of the temperature distribution data and the material state spectrum data; the at least two independent heating partitions include a center main heating zone and at least two edge compensation heating zones.

[0029] Specifically, the trigger terminals of the infrared thermal imager (such as a non-cooled micro-thermal detector) and the online near-infrared spectrometer (such as an InGaAs array detector) are connected to the same digital output port of the collaborative control unit. The control unit sends a synchronization pulse signal, and both capture a frame of temperature image and a spectrum curve at the same millisecond level.

[0030] Specifically, in step S2, the spatial domain analysis of the temperature distribution data includes: calculating the standard deviation, coefficient of variation or range of the cross-sectional temperature distribution as a temperature uniformity characteristic parameter; and / or, calculating the average temperature difference between the pre-set center region and the edge region in the cross-section as a temperature uniformity characteristic parameter.

[0031] Specifically, in step S3, the frequency domain or characteristic wavelength analysis of the material state spectrum data includes: identifying the characteristic absorption peak corresponding to the chemical bond of the yarn material, and calculating the intensity, peak area or intensity ratio relative to the reference peak of the characteristic absorption peak as the material state characteristic data; and / or, calculating the derivative, slope or curvature of the material state spectrum data in a specific waveband as the material state characteristic data.

[0032] Spatial domain analysis example: The control system divides the temperature image into analysis regions corresponding to the physical heating zones. For example, there is a central analysis region corresponding to the central main heating zone (accounting for about 60% of the cross-sectional width) and an edge analysis region corresponding to the edge compensation heating zone (each accounting for about 20%). Calculate the standard deviation σ of the temperature values ​​of all pixels in the entire image, and simultaneously calculate the average temperature of the edge regions. Average temperature in the central area The difference ΔT. σ and ΔT are used as characteristic parameters of temperature uniformity; Example of frequency domain analysis: For polyester filament tow, the focus is on the ester group C=O bond at approximately 1720 cm⁻¹. -1 Characteristic absorption peak at the location; Analyze the material's state spectral data and calculate the peak intensity of the absorption peak. And calculate its position at 1700cm. -1 -1740cm -1 Integral area within the range ; Will and As material state characteristic data.

[0033] Specifically, in the subsequent calculation of composite state indices, the following can be selected: or One of them can be used as a representative feature data, or a comprehensive index that includes both can be constructed.

[0034] Specifically, in step S4, determining the composite state index includes: normalizing the temperature uniformity characteristic parameters and material state characteristic data respectively, and then weighting and fusing them according to preset weighting coefficients to generate a scalar form composite state index.

[0035] Specifically, the feature parameters are first normalized: ,

[0036] in The target peak intensity under ideal conditions; Then, weighted fusion is performed: composite state index

[0037] Where w1 and w2 are preset weights, such as w1=0.6 and w2=0.4. The closer the F value is to 1, the better the state.

[0038] Please see Figure 2 As shown, Figure 2 A flowchart illustrating the steps for generating a coordinated power adjustment command in an embodiment of the present invention.

[0039] Specifically, in step S5, generating the cooperative power adjustment instruction comprises the following sub-steps: Step S51, input the composite state index into the heat transfer model to predict the future trend of the filament state under different power adjustment schemes; Step S52, according to the composite state index, query the rule base to obtain the power adjustment strategy based on expert experience; Step S53, using a preset arbitration algorithm, comprehensively consider the future trend and the power adjustment strategy to generate the final cooperative power adjustment instruction.

[0040] Specifically, the heat transfer model is established based on the law of conservation of energy, and its control equation can be expressed as:

[0041] Wherein, ρ is the density of the filament material, is the specific heat capacity of the filament material, k is the effective thermal conductivity of the filament material in the heating state, T is the temperature field function, t is the time, x and y are the spatial coordinates on the cross section of the filament; ∂T / ∂t represents the rate of change of temperature T with time t, i.e. the instantaneous rate of heating or cooling; ∂²T / ∂x² and ∂²T / ∂y² respectively represent the second-order spatial derivatives of temperature T in the x and y directions, which together describe the intensity of heat diffusion within the cross section of the filament due to temperature non-uniformity; Q(x,y,t) is a spatial heat source function corresponding to the power distribution of each heating sub-zone. The model is discretely solved on the cross section grid using finite difference method or finite element method. The physical property parameters of the density ρ, specific heat capacity , thermal conductivity k and infrared absorption rate, etc. can be obtained through standard thermal property tests for different filament specifications and input during model initialization.

[0042] It can be understood that the density ρ, specific heat capacity c p , effective thermal conductivity k and other thermal property parameters of the filament material can be pre-determined through standard thermal property tests (such as differential scanning calorimetry, thermal conductivity meter measurement) for different materials (such as polyester, nylon, polypropylene) and specifications (such as fineness, cross-sectional shape), and a material parameter library is established and stored in the control system. In actual application, after the operator selects the product specification, the system automatically calls the corresponding material parameters for model initialization.

[0043] Specifically, the confidence C of the composite state index F can be evaluated by one or a combination of the following methods: Based on signal integrity evaluation: Check the signal-to-noise ratio (SNR) of the infrared thermal imaging signal and the spectral signal respectively; If both SNRs are higher than a preset threshold, in this embodiment, the preset threshold is 30dB, then set a higher confidence (e.g. C=0.9); If either SNR is lower than the threshold, then reduce the confidence accordingly.

[0044] Specifically, as an implementable quantitative example, the confidence C can be calculated as follows:

[0045] wherein, and are the signal-to-noise ratios of the infrared thermal imaging signal and the spectral signal respectively, is a preset signal-to-noise ratio threshold (e.g. 30dB), is the moving average of the composite state indicator F in the past N sampling periods, α, β, γ are weighting coefficients and satisfy α+β+γ=1, and δ is a normalization parameter. When the signal quality is high and the data is stable, C tends to 1. In this embodiment, the weighting coefficients α, β, γ can be preset according to the relative importance of the infrared thermal imaging and spectral analysis in state evaluation, for example, they can be set to 0.4, 0.4, and 0.2 respectively. The normalization parameter δ is used to adjust the sensitivity of F value fluctuation, which can be set according to the fluctuation range of F value in historical data, for example, it can be set to 0.1. These parameters can be determined by experimental calibration during system debugging.

[0046] Evaluation based on data consistency: Calculate the deviation of the current F value from the sliding average of F values in the past period. The smaller the deviation, the higher the confidence.

[0047] The confidence C can be quantified as a value between 0 and 1, and used as an explicit input of the arbitration algorithm.

[0048] Specifically, the rule base can include the following rules: Rule 1: Based on the result of F<0.85 and ΔT>5℃, determine to significantly increase the power of the edge zone and slightly increase the power of the center zone.

[0049] Rule 2: Based on the result of F>0.95, determine to reduce all partition powers by 3% to save energy; Rule 3: Based on the result of ΔT<0 and (Abs_peak(t)-Abs_peak(t-Δt)) / Δt<-R_th, determine to increase the power of the center zone, maintain or slightly reduce the power of the edge zone, and issue a warning that the filament center area may be overheating; Wherein, Abs peak (t) represents the characteristic absorption peak intensity at the current time, Abs peak (t-Δt) represents the intensity at the previous sampling time, Δt is the sampling interval time, -R_th is a preset negative change rate threshold (for example, -0.05 / ms). The change rate of the material state characteristic data can be calculated and judged in this way. The rule base can be edited and expanded through a configuration file.

[0050] Specifically, the arbitration algorithm is used to integrate the prediction results of the heat transfer model and the reasoning strategy of the rule base. One embodiment thereof is configured to dynamically adjust the weights of the two according to the confidence C of the composite state indicator F: set a confidence threshold C_th, in this embodiment, C_th=0.8, and the algorithm performs the following logic: Based on the result of C≥C_th, the model prediction weight W_model=0.7, and the rule reasoning weight W_rule=0.3; Based on the result of C<C_th, the model prediction weight W_model=0.3, and the rule reasoning weight W_rule=0.7; Final instruction=W_model×model prediction result+W_rule×rule reasoning result; The confidence threshold C_th and the corresponding weight distribution (W_model, W_rule) can be calibrated and determined through offline simulation, historical data backtesting or on-site debugging, with system control stability and response speed as the optimization objective.

[0051] This embodiment shows the dynamic weight adjustment based on the confidence partition interval. It can be understood that the weight can also be a continuous function of the confidence C, realizing smoother dynamic adjustment.

[0052] Specifically, the set of process parameters is also associated with the suggested value of the false twister process parameter linked with the initial heating parameter.

[0053] Specifically, in step S7, the process of optimizing the control parameters includes: Step S71, record the composite state indicator, the executed cooperative power adjustment instruction and the corresponding product quality data in the historical production data; Step S72, analyze the historical production data through a machine learning algorithm to optimize the parameters of the heat transfer model, the rules in the rule base or the weight coefficient in step S4.

[0054] Specifically, the system generates an adjustment scheme for the optimization suggestions (such as updated model parameters, rule thresholds or fusion weights) generated by machine learning analysis, and after being confirmed by an operator or automatically reviewed by the system, it is updated online or offline to the corresponding heat transfer model, rule base or fusion weight configuration, realizing closed-loop self-optimization of the control parameters.

[0055] Specifically, the machine learning algorithm can be selected as Gradient Boosting Decision Tree (GBDT) or neural network algorithm. The optimization process is as follows: the temperature uniformity feature parameters, material state feature data and executed power adjustment instructions in the historical data are taken as features (X), and the corresponding product quality data (such as dyeing uniformity score) is taken as label (Y), and a prediction model is trained. By analyzing the importance of each feature in the prediction model, and finding the combination of X and instructions that optimizes the prediction quality Y, the adjustment suggestions for the heat transfer model parameters, rule triggering threshold or fusion weight (w1, w2) are reversely deduced.

[0056] Specifically, the system continuously runs and records the feature parameters-F value-adjustment instruction-finished yarn physical properties (such as breaking strength, dyeing uniformity grade) of each batch. Periodically (such as every 100 batches of production), these data are used to train a regression model, which takes feature parameters and adjustment instructions as input and product quality prediction as output. By adjusting the model parameters through the optimization algorithm, the prediction quality is optimized, and then the improvement suggestions for the heat transfer model, rule base or fusion weight w1 / w2 found in this process are updated to the system.

[0057] So far, the technical solutions of the present application have been described in combination with the preferred embodiments shown in the drawings, but it is easy for those skilled in the art to understand that the protection scope of the present application is obviously not limited to these specific embodiments. Those skilled in the art can make equivalent changes or replacements to the related technical features without departing from the principles of the present application, and the technical solutions after these changes or replacements will fall within the protection scope of the present application.

[0058] The above description is only the preferred embodiments of the present application and is not used to limit the present application; for those skilled in the art, the present application can have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A false twist texturing machine infrared heat chamber cooperative heating control method, characterized by, The method comprises the following steps: Step S1, setting initial power of each heating partition according to product specification of to-be-processed tow; Step S2, collecting cross-section temperature distribution data of tow passing through the heat box, and performing spatial domain analysis on the temperature distribution data to extract temperature uniformity characteristic parameters; Step S3, collecting material state spectrum data of the material passing through the heat box, and performing frequency domain or characteristic wavelength analysis on the material state spectrum data to extract material state characteristic data; Step S4, determining a composite state index based on the temperature uniformity characteristic parameters and the material state characteristic data; Step S5, generating a coordinated power adjustment instruction for at least two independent heating partitions in the heat box based on the composite state index, a heat transfer model and a rule base; Step S6, independently adjusting the heating power of each heating partition according to the coordinated power adjustment instruction; Step S7, optimizing control parameters based on historical production data.

2. The false twist texturing machine infrared heat oven cooperative heating control method according to claim 1, characterized in that, In the step S1, initial heating parameters corresponding to the current product specification are called from a pre-stored process parameter set to set the initial power of each heating partition; wherein the process parameter set at least stores different product specifications and their corresponding initial heating parameters.

3. The false twist texturing machine infrared heat box coordinated heating control method according to claim 2, characterized in that, The data collection of steps S2 and S3 is performed by a synchronous triggered infrared thermal imaging device and an online spectrometer to ensure the time consistency of the temperature distribution data and the material state spectrum data; the at least two independent heating partitions include a central main heating zone and at least two edge compensation heating zones.

4. The false twist texturing machine infrared heat box coordinated heating control method according to claim 3, characterized in that, In the step S2, the spatial domain analysis on the temperature distribution data includes: calculating the standard deviation, coefficient of variation or range of the cross-section temperature distribution as the temperature uniformity characteristic parameters; and / or, calculating the average temperature difference between the preset center region and the edge region in the cross-section as the temperature uniformity characteristic parameters.

5. The false twist texturing machine infrared heat box coordinated heating control method according to claim 4, characterized in that, In the step S3, the frequency domain or characteristic wavelength analysis on the material state spectrum data includes: identifying a characteristic absorption peak corresponding to the chemical bond of the tow material, and calculating the intensity, peak area or intensity ratio relative to the reference peak of the characteristic absorption peak as the material state characteristic data; and / or, calculating the derivative, slope or curvature of the material state spectrum data in a specific waveband as the material state characteristic data.

6. The false twist texturing machine infrared heat box coordinated heating control method according to claim 5, characterized in that, In the step S4, the determination of the composite state index includes: normalizing the temperature uniformity characteristic parameters and the material state characteristic data respectively, and generating the composite state index in scalar form by weighted fusion according to a pre-set weight coefficient.

7. The false twist texturing machine infrared heat box coordinated heating control method according to claim 6, characterized in that, In the step S5, the generation of the coordinated power adjustment instruction includes the following sub-steps: Step S51, inputting the composite state index into the heat transfer model to predict the future change trend of the tow state under different power adjustment schemes; Step S52, querying the rule base according to the composite state index to obtain a power adjustment strategy based on expert experience; Step S53, generating the final coordinated power adjustment instruction by comprehensively using a pre-set arbitration algorithm and the future change trend and the power adjustment strategy.

8. The false twist texturing machine infrared heat box coordinated heating control method according to claim 7, characterized in that, The arbitration algorithm is configured to dynamically adjust the weight proportion of the prediction result of the heat transfer model and the inference result of the rule base in the final instruction according to the confidence of the composite state indicator or the change rate of the material state feature data.

9. The false twist texturing machine infrared heat box coordinated heating control method according to claim 8, characterized in that, The set of process parameters further stores a false twister process parameter suggestion value linked with the initial heating parameter.

10. The false twist texturing machine infrared heat box coordinated heating control method according to claim 9, characterized in that, In the step S7, the process of optimizing the control parameters includes: Step S71, record the composite state indicators, the executed cooperative power adjustment instructions and the corresponding product quality data in the historical production data; Step S72, analyze the historical production data by a machine learning algorithm to optimize the parameters of the heat transfer model, the rules in the rule base or the weight coefficients in step S4.

Images