A temperature intelligent control method for edible glue drying process

Abstract

The present invention relates to the technical field of edible glue drying, and in particular to a temperature intelligent control method for edible glue drying process. The method comprises: calculating the multivariate distance metric of the matching data point pairs in two historical drying sequences; calculating the distance between the two historical drying sequences using the multivariate distance metric of the matching data point pairs and their corresponding weights; clustering based on the distance between the historical drying sequences, and recording each category as a drying mode; obtaining the optimal PID parameters of each drying mode according to the PID parameters corresponding to the historical drying sequences in the drying mode; obtaining the similarity between the drying sequence before each parameter update time point in the current drying process and the cluster center corresponding to each drying mode; adjusting the PID parameters according to the similarity and the optimal PID parameters of each drying mode. The present invention can realize the adaptive control of the PID parameters in the edible glue drying process and improve the drying quality.

Description

Technical Field

[0001] The invention relates to the technical field of edible glue drying, and in particular to a temperature intelligent control method used in the edible glue drying process. Background Art

[0002] Edible glue is an important additive widely used in the food industry. Its application range includes but is not limited to thickeners, stabilizers and gelling agents. In the production process of edible glue, drying is a key step that directly affects the quality, functional properties and production efficiency of the product. In recent years, adaptive control technology has been gradually applied to the drying process of food glue. By real-time monitoring and adjustment of drying parameters, the uniformity and stability of the drying process can be improved.

[0003] The existing technology improves the PID control method by combining the real-time data collected by the temperature and humidity sensor with the expert experience to set parameters to improve the temperature control accuracy. However, this type of method still has difficulty in effectively handling the drying process with obvious nonlinear and time-varying characteristics, and has the problem of insufficient adaptability. Different batches of raw materials may have differences in composition and water content, and the fixed temperature curve is difficult to adapt to these changes, which can easily lead to unstable drying effects. Although the traditional PID control method is widely used in temperature control, it is insufficient in dealing with dynamic changes in the drying process and cannot fully consider the time delay and nonlinear characteristics of the drying process. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a temperature intelligent control method for edible glue drying process, and the technical scheme adopted is as follows:

[0005] An embodiment of the present invention provides a method for intelligently controlling temperature during the drying process of edible glue, the method comprising:

[0006] Obtain a historical drying sequence, the historical drying sequence including the temperature and humidity of a drying process; calculate the temperature change rate and humidity change rate corresponding to each temperature and humidity in the historical drying sequence; calculate a multivariate distance metric for matching data point pairs in two historical drying sequences based on the temperature change rate and humidity change rate;

[0007] Calculate the weight corresponding to the multivariate distance metric of the matching data point pair according to the temperature difference and humidity difference in the matching data point pair; calculate the distance between two historical drying sequences using the multivariate distance metric of the matching data point pair and its corresponding weight;

[0008] Clustering is performed based on the distance between the historical drying sequences, and each category is recorded as a drying mode; and the optimal PID parameters of each drying mode are obtained according to the PID parameters corresponding to the historical drying sequences in the drying mode;

[0009] The similarity between the drying sequence before each parameter update time point in the current drying process and the cluster center corresponding to each drying mode is obtained; and the PID parameters of the current drying process are adjusted according to the similarity and the optimal PID parameters of each drying mode.

[0010] Preferably, a historical drying sequence is obtained, where the historical drying sequence includes the temperature and humidity of a drying process, including:

[0011] The temperature data and humidity data of the drying process are collected by using a temperature sensor and a humidity sensor respectively, and the sampling interval is a preset sampling interval; the temperature data and humidity data of a drying process constitute a historical drying sequence corresponding to the drying process.

[0012] Preferably, calculating the temperature change rate and humidity change rate corresponding to each temperature and humidity in the historical drying sequence respectively includes:

[0013] Obtain the difference between the temperature at a later moment and the temperature at a previous moment in the historical drying sequence, and the ratio of the difference to the time interval between the two moments is the temperature change rate corresponding to the temperature at the later moment;

[0014] The difference between the humidity at a later moment and the humidity at a previous moment in the historical drying sequence is obtained, and the ratio of the difference to the time interval between the two moments is the humidity change rate corresponding to the humidity at the later moment.

[0015] Preferably, the multivariate distance metric for matching pairs of data points in two historical drying sequences is:

[0016] ,

[0017] in, represents a multivariate distance measure for pairs of matching data points, and They represent the data point of the historical drying sequence X at time i and the data point of the historical drying sequence Y at time j respectively; and Respectively represent the temperature and humidity of the historical drying sequence X at time i; and represent the temperature and humidity of the historical drying sequence Y at time j respectively; and They represent the relative rate of change of temperature and the relative rate of change of humidity respectively;

[0018] The relative change rates of temperature and humidity are:

[0019] ,

[0020] ,

[0021] in, and represents the temperature change rate and humidity change rate at time i in the historical drying sequence X; and They respectively represent the temperature change rate and humidity change rate at time j in the historical drying sequence Y; | | represents the absolute value symbol.

[0022] Preferably, the weights corresponding to the multivariate distance metric for the matching data point pairs are:

[0023] ,

[0024] in, Represents the weights corresponding to the multivariate distance measure for matching pairs of data points; and Respectively represent the temperature and humidity of the historical drying sequence X at time i; and They represent the temperature and humidity of the historical drying sequence Y at time j respectively; k represents the adjustment parameter of the sensitivity function; σ represents the adjustment parameter of the time distance; Represents a series of index pairs in the DTW algorithm The alignment paths of the components; represents the exponential function with the natural constant e as the base; N represents the number of all matching data point pairs in the DTW metric.

[0025] Preferably, the distance between two historical drying sequences is:

[0026] ,

[0027] in, Represents the distance between historical drying sequence X and historical drying sequence Y; Represents the weights corresponding to the multivariate distance measure for matching pairs of data points; represents a multivariate distance measure for pairs of matching data points, and They represent the data point of the historical drying sequence X at time i and the data point of the historical drying sequence Y at time j respectively; Represents a series of index pairs in the DTW algorithm The alignment paths of the components; Represents the set of all alignment paths.

[0028] Preferably, the optimal PID parameters of each drying mode are obtained according to the PID parameters corresponding to the historical drying sequences in the drying mode, including:

[0029] The optimal PID parameters of each drying mode include the optimal proportional gain parameter, the optimal integral gain parameter and the optimal differential gain parameter;

[0030] The average of the proportional gain parameters in the PID parameters corresponding to the preset number of historical drying sequences with the highest drying effect scores in each drying mode is the optimal proportional gain parameter of the drying mode;

[0031] The average of the integral gain parameters in the PID parameters corresponding to the preset number of historical drying sequences with the highest drying effect scores in each drying mode is the optimal integral gain parameter of the drying mode;

[0032] The average of the differential gain parameters in the PID parameters corresponding to a preset number of historical drying sequences with the highest drying effect scores in each drying mode is the optimal differential gain parameter of the drying mode.

[0033] Preferably, obtaining the similarity between the drying sequence before each parameter update time point in the current drying process and the cluster center corresponding to each drying mode includes:

[0034] In the current drying process, parameter update time points are set at preset intervals; the temperature and humidity data before each parameter update time point are obtained to form a drying sequence; the similarity calculation formula is:

[0035] ,

[0036] in, Indicates the drying sequence corresponding to the gth parameter update time point and the gth parameter update time point of the current drying process. The similarity of the cluster centers of the drying modes; is the drying sequence corresponding to the current drying process from the beginning to the g-th parameter update time point; Indicates The cluster centers of the drying modes; exp( ) represents the exponential function with the natural constant as the base, and DTW( ) represents the dynamic time warping algorithm.

[0037] Preferably, the PID parameters of the current drying process are adjusted according to the similarity and the optimal PID parameters of each drying mode, including:

[0038] Calculate the updated proportional gain parameter, updated integral gain parameter and updated differential gain parameter corresponding to the g-th parameter update time point in the current drying process; the calculation formulas are , and ;

[0039] in, , and They respectively represent the updated proportional gain parameter, updated integral gain parameter and updated differential gain parameter corresponding to the g-th parameter update time point in the current drying process; Indicates the drying sequence corresponding to the gth parameter update time point and the gth parameter update time point of the current drying process. The similarity of the cluster centers of the drying modes; , and Respectively represent The optimal proportional gain parameter, optimal integral gain parameter and optimal differential gain parameter of each drying mode; M represents the number of drying modes;

[0040] The PID parameters between the g-th parameter update time point and the g+1-th parameter update time point are adjusted by updating the proportional gain parameter, updating the integral gain parameter and updating the differential gain parameter.

[0041] Preferably, after obtaining the historical drying sequence, the method further includes:

[0042] Experts are used to score the drying effect of each historical drying sequence to obtain the drying effect score of each historical drying sequence.

[0043] The embodiments of the present invention have at least the following beneficial effects: the present invention obtains a historical drying sequence, and then calculates the temperature change rate and humidity change rate corresponding to each temperature and humidity respectively, and calculates a multivariate distance metric for matching data point pairs in two historical drying sequences based on the temperature change rate and the humidity change rate, and optimizes the determination of the multivariate distance metric to ensure that when calculating the distance between matching data point pairs in two historical drying sequences, a higher weight is given to variables with a high relative change rate, so that the obtained multivariate distance metric is closer to the actual drying process, thereby improving the accuracy of subsequent analysis;

[0044] The weights corresponding to the multivariate distance metric of the matching data point pairs are calculated according to the temperature difference and humidity difference in the matching data point pairs; the distance between two historical drying sequences is calculated using the multivariate distance metric of the matching data point pairs and their corresponding weights, and the influence of the mutation points and time proximity of temperature and humidity are comprehensively considered, thereby improving the robustness of the historical drying distance calculation;

[0045] Furthermore, multiple drying modes are obtained through clustering, and the optimal PID parameters of each drying mode are calculated. At the same time, the similarity between the drying sequence before each parameter update time point in the current drying process and the cluster center corresponding to each drying mode is obtained. The PID parameters of the current drying process are adjusted according to the similarity and the optimal PID parameters of each drying mode. This not only realizes the adaptive control of PID parameters, but also improves the accuracy of temperature control. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 A method flow chart of a method for intelligent temperature control in the drying process of edible glue provided by the present invention. DETAILED DESCRIPTION

[0048] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the temperature intelligent control method for edible glue drying process proposed by the present invention, its specific implementation, structure, characteristics and effects as follows. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0049] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0050] The specific scheme of the temperature intelligent control method for the edible glue drying process provided by the present invention is described in detail below with reference to the accompanying drawings.

[0051] Example:

[0052] The main application scenarios of the present invention are: in the drying process of edible glue, changes in temperature and humidity play a decisive role in the quality of the final product. Due to differences in raw material batches, changes in external environmental conditions and other factors, the drying processes of different batches may present different temperature and humidity change patterns. This diversity makes it difficult to adapt to all situations using a single control strategy, which can easily lead to unstable product quality and energy waste. How to accurately identify the temperature change pattern during the drying process of edible glue, and dynamically adjust the control strategy accordingly to improve the drying efficiency and consistency of product quality. The present invention performs feature extraction and distance metric optimization for different historical drying sequences during the drying process of food glue, thereby completing k-means clustering analysis and obtaining adaptive PID control parameters to achieve the purpose of intelligent temperature control.

[0053] See also Figure 1 , which shows a flow chart of a temperature intelligent control method for edible glue drying process provided by an embodiment of the present invention, the method comprising the following steps:

[0054] Step S1, obtaining a historical drying sequence, the historical drying sequence including the temperature and humidity of a drying process; calculating the temperature change rate and humidity change rate corresponding to each temperature and humidity in the historical drying sequence; and calculating the multivariate distance metric of matching data point pairs in two historical drying sequences based on the temperature change rate and the humidity change rate.

[0055] In the process of drying edible glue, the temperature and humidity changes in different drying processes are different. It is difficult to adapt to all situations to use a single PID parameter (proportional gain parameter P, integral gain parameter I and differential gain parameter D) control strategy to control the temperature in the drying process, which affects the quality of drying. At the same time, in the drying process, not only the temperature changes need to be considered, but also the humidity changes.

[0056] The temperature and humidity data in the drying process are collected using a temperature sensor and a humidity sensor. The sampling interval is a preset sampling interval. Preferably, in the embodiment of the present invention, the preset sampling interval is 5 seconds. Thus, historical data is obtained to form multiple historical drying sequences, wherein one drying process corresponds to one historical drying sequence. When recording data, the timestamp of the starting point of drying is 0, and subsequent data is recorded with a timestamp according to the set sampling interval. The data format in each historical drying sequence X is .in Drying Sequences for History The temperature data of the i-th data point in, Drying Sequences for History The humidity data of the i-th data point in the drying sequence can also be expressed as the data point at time i in the historical drying sequence. The data point at time i includes temperature data and humidity data, which can be expressed as , t is the timestamp. The collected data is preprocessed, including removing outliers and smoothing.

[0057] After obtaining the historical drying sequence, it is also necessary to use experts to score each historical drying sequence according to the quality of the edible glue after drying to obtain the drying effect score of each historical drying sequence. Then the historical drying sequence and its corresponding drying effect score are combined into a data set.

[0058] In the present invention, it is necessary to identify several typical drying modes and obtain an adaptive control strategy. Traditional time series analysis methods are often difficult to handle some special cases encountered in the drying process, such as: different batches have different drying times, resulting in inconsistent curve lengths. Time delays or advances occur in the drying process, making similar processes misaligned on the time axis. Taking these actual situations into account, the dynamic time warping (DTW) algorithm is selected as the basis and optimized to better adapt to the characteristics of the edible glue drying process.

[0059] According to the above requirements, the DTW algorithm is optimized and a multivariate distance metric based on relative rate of change is introduced. The purpose of this is to consider the changes in temperature and humidity at the same time and dynamically adjust their importance according to the different stages of the drying process. For example, in the early stage of drying, the change in humidity may be more critical, while in the later stage, the precise control of temperature may be more important. With this improvement, the similarity between different drying processes can be evaluated more accurately.

[0060] Furthermore, the temperature change rate and humidity change rate corresponding to each temperature and humidity in each historical drying sequence are calculated. The traditional DTW algorithm is mainly used for univariate time series comparison. In order to consider the changes in temperature and humidity at the same time, a multivariate distance metric DTW algorithm based on relative change rate is introduced.

[0061] During the drying process of edible glue, the changes in temperature and humidity are not uniform. For example, in the initial heating stage, the temperature changes faster, while the humidity changes relatively slowly; in the constant temperature drying stage, the temperature changes less, while the humidity drops faster. When entering the heating stage, the temperature change rate will become larger, which means that the temperature change is more important in this stage, and the DTW algorithm will focus on the temperature change. Similarly, in the critical stage of the formation of the colloidal structure, the precise control of humidity is crucial. At this time, the humidity change rate will change significantly, and the algorithm can better capture this process. The weights of the two variables of temperature and humidity in the stage are controlled by the relative change rate. For multivariate distance measurement, since the collected time series contains two key variables, temperature and humidity. In order to consider the changes in temperature and humidity at the same time, the distance measurement of the selected two data points is adjusted by the relative change rates of temperature and humidity in different stages. On this basis, the calculation of DTW between different time series is completed.

[0062] Calculate the temperature change rate and humidity change rate corresponding to each temperature and humidity in the historical drying sequence. The temperature change rate is calculated as follows: obtain the difference between the temperature at the next moment and the temperature at the previous moment in the historical drying sequence. The ratio of the difference to the time interval between the two moments is the temperature change rate corresponding to the temperature at the next moment. The calculation formula is: ; The calculation of humidity change rate is: obtain the difference between the humidity at the next moment and the humidity at the previous moment in the historical drying sequence, and the ratio of the difference to the time interval between the two moments is the humidity change rate corresponding to the humidity at the next moment. The calculation formula is: .in Represents the time interval between two moments.

[0063] When the DTW algorithm is used to calculate the distance between two time series, multiple alignment paths are obtained. Each alignment path includes multiple matching data point pairs. Each matching data point pair includes two matching data points in the two time series, which are also two matching data points in the two historical drying sequences in the present invention. One data point includes the temperature and humidity at the same time, thereby calculating the multivariate distance metric of the matching data point pair. , the calculation formula is:

[0064] ,

[0065] in, represents a multivariate distance measure for pairs of matching data points, and They represent the data point of the historical drying sequence X at time i and the data point of the historical drying sequence Y at time j respectively; and Respectively represent the temperature and humidity of the historical drying sequence X at time i; and represent the temperature and humidity of the historical drying sequence Y at time j respectively; and They represent the relative rate of change of temperature and the relative rate of change of humidity respectively;

[0066] Among them, the relative change rate of temperature and the relative change rate of humidity are:

[0067] ,

[0068] ,

[0069] in, and represents the temperature change rate and humidity change rate at time i in the historical drying sequence X; and They respectively represent the temperature change rate and humidity change rate at time j in the historical drying sequence Y; | | represents the absolute value symbol.

[0070] When calculating the multivariate distance metric for matching data point pairs, the relative change rates of temperature and humidity are introduced to ensure that variables with high relative change rates are given higher weights, thereby making the algorithm's focus more in line with the actual situation of the drying process.

[0071] Step S2, calculating the weight corresponding to the multivariate distance metric of the matching data point pair according to the temperature difference and humidity difference in the matching data point pair; and calculating the distance between the two historical drying sequences using the multivariate distance metric of the matching data point pair and its corresponding weight.

[0072] In the above step S1, the adaptive weight optimization factors of the relative rate of change of temperature and the relative rate of change of humidity are obtained to calculate the multivariate distance metric, which improves the accuracy of the DTW algorithm when processing the data of the edible glue drying process. However, there are still detailed problems to be solved in this optimization: in the drying process, the changes in some stages are more important than those in other stages. However, the above processing does not take into account the changes in the drying effect at different stages.

[0073] For example, in the rapid dehydration stage at the beginning of drying, the humidity value will change greatly, while the temperature value will rise slowly. Then, when performing DTW calculations in a sequence, the distance calculation between two matching data points will increase the humidity weight in the distance calculation of the current data point pair due to the introduction of the relative change rate control weight. The weight calculation is only calculated based on the change rate between the data point pairs. The relative change rate simply amplifies the variable with a larger change rate in the two matching data point pairs. It does not capture the comprehensive difference between the two sequences at a specific moment.

[0074] Specifically, it is necessary to perform secondary optimization on the distance between pairs of data points through a dynamic weight optimization factor, which comprehensively considers the influence of the mutation points of temperature and humidity and the degree of time proximity.

[0075] In the optimization of step S1, the variables with high change rates in the matching data point pairs are highlighted through the influence of the change rate. It is also necessary to introduce a dynamic weight optimization factor to optimize the contribution of each matching data point pair to the similarity between sequences. Let those matching data point pairs with high change rates and close time have greater weights. Simply put, the optimization in step S1 selects the more important variables between the data point pairs, and the secondary optimization in step S2 amplifies the similarity contribution of data points with high change rates and close time. In the subsequent clustering process, drying sequences with similar parameters are easily clustered into one category. Therefore, it is necessary to calculate the weights corresponding to the multivariate distance metric of the matching data point pairs based on the temperature difference and humidity difference in the matching data point pairs. The calculation formula is:

[0076] ,

[0077] in, Represents the weights corresponding to the multivariate distance measure for matching pairs of data points; and Respectively represent the temperature and humidity of the historical drying sequence X at time i; and They represent the temperature and humidity of the historical drying sequence Y at time j respectively; k represents the adjustment parameter of the sensitivity function, which controls the steepness of the sensitivity function, and its initial value is set to 0.1; σ represents the adjustment parameter of the time distance, which controls the width of the time weight function, reflecting the influence of the proximity of the time points on the weight, and its initial value is set to 5. Grid search optimization can be performed according to the processed data; Represents a series of index pairs in the DTW algorithm The alignment paths of the components; represents an exponential function with the natural constant e as the base; N represents the number of all matching data point pairs in the DTW metric, that is, the number of all aligned data points in the DTW metric.

[0078] In the above formula, Reflects the impact of temperature and humidity differences on weights. The greater the difference between the data, the greater the impact. When the difference is close to or exceeds the threshold, it will gain a greater weight, otherwise it will reduce the weight. Highlight the matching data point pairs that need to be focused on (high rate of change and large difference);

[0079] is the threshold parameter, which determines at what difference the weights begin to change significantly. Indicates the number of aligned data points in the two selected series;

[0080] is a time weight function, so that the closer the points in two historical drying sequences are to each other in time, the greater the weight is.

[0081] Furthermore, the distance between the two historical drying sequences is calculated using the multivariate distance metric of the matching data point pairs and their corresponding weights, that is, the distance between the two historical drying sequences is calculated using the DTW algorithm. The calculation formula is:

[0082] ,

[0083] in, Represents the distance between historical drying sequence X and historical drying sequence Y; Represents the weights corresponding to the multivariate distance measure for matching pairs of data points; represents a multivariate distance measure for pairs of matching data points, and They represent the data point of the historical drying sequence X at time i and the data point of the historical drying sequence Y at time j respectively; Represents a series of index pairs in the DTW algorithm The alignment paths of the components; Represents the set of all alignment paths.

[0084] Finally, the distance between every two historical drying sequences can be calculated for subsequent analysis.

[0085] Step S3, clustering is performed based on the distance between the historical drying sequences, and each category is recorded as a drying mode; and the optimal PID parameters of each drying mode are obtained according to the PID parameters corresponding to the historical drying sequences in the drying mode.

[0086] After obtaining the distances between all the above-mentioned historical drying sequences, k-means clustering analysis is performed on all the historical drying sequences. Since k-means hierarchical clustering is an existing method, it will not be described here. It should be noted that the clustering results are analyzed to determine the optimal number of clusters. K-means clustering needs to use the elbow rule to determine the number of clusters. At the same time, the silhouette coefficient and Calinski-Harabasz index can also be calculated according to existing analysis methods. The closer the former is to 1, the better the effect, and the larger the value of the latter, the better the effect. The above methods are all existing methods and can be selected according to specific accuracy requirements. The formula description is not given here.

[0087] Thus, the clustering category can be obtained, and the clustering category can be expressed as a drying mode. After obtaining the above clustering results, the PID parameter optimization based on the drying mode can be performed according to the clustering results. For the drying mode calculated above, the drying effect scores of the sequences within the cluster are sorted and the preset number of historical drying sequences with the highest scores are selected to extract their historical PID parameters. The value of the preset number can be determined according to the amount of data. Preferably, the value in the present invention is 20, because the drying sequence of the center point of the cluster does not necessarily meet all drying conditions.

[0088] The average of the proportional gain parameters in the PID parameters corresponding to the preset number of historical drying sequences with the highest drying effect scores in each drying mode is the optimal proportional gain parameter of the drying mode, and the calculation formula is: , represents the optimal proportional gain parameter of the mth drying mode, represents the proportional gain parameter of the nth historical drying sequence in the mth drying mode, and U represents the number of historical drying sequences in the mth drying mode;

[0089] The average of the integral gain parameters in the PID parameters corresponding to the preset number of historical drying sequences with the highest drying effect scores in each drying mode is the optimal integral gain parameter of the drying mode, and the calculation formula is: , represents the optimal integral gain parameter of the mth drying mode, represents the integral gain parameter of the nth historical drying sequence in the mth drying mode, and U represents the number of historical drying sequences in the mth drying mode;

[0090] The average of the differential gain parameters in the PID parameters corresponding to the preset number of historical drying sequences with the highest drying effect scores in each drying mode is the optimal differential gain parameter of the drying mode, and the calculation formula is: , represents the optimal differential gain parameter of the mth drying mode, represents the differential gain parameter of the nth historical drying sequence in the mth drying mode, and U represents the number of historical drying sequences in the mth drying mode.

[0091] The optimal PID parameters of each drying mode include an optimal proportional gain parameter, an optimal integral gain parameter and an optimal differential gain parameter.

[0092] Step S4, obtaining the similarity between the drying sequence before each parameter update time point in the current drying process and the cluster center corresponding to each drying mode; adjusting the PID parameters of the current drying process according to the similarity and the optimal PID parameters of each drying mode.

[0093] In the current drying process, the PID parameters need to be updated in real time, so the parameter update time point is set. Preferably, a parameter update time point is set at a preset interval in the present invention. Preferably, the preset interval is 300 seconds. The implementer can adjust the preset interval according to the computing power, sampling accuracy and actual control conditions of the actual equipment. The temperature and humidity data before each parameter update time point are obtained to form a drying sequence. For example, the temperature and humidity data of 0-300 seconds are obtained at the first parameter update time point to form a drying sequence, and the temperature and humidity data of 0-600 seconds are obtained at the second parameter update time point to form a drying sequence. That is, the PID parameters are updated every 300 seconds in the current drying process.

[0094] Calculate the similarity between the drying sequence corresponding to each parameter update point in the current drying process and the cluster center of each drying mode, and the closeness between the current drying process and the mth drying mode. The formula is as follows:

[0095] ,

[0096] in, Indicates the drying sequence corresponding to the gth parameter update time point and the gth parameter update time point of the current drying process. The similarity of the cluster centers of the drying modes; is the drying sequence corresponding to the current drying process from the beginning to the g-th parameter update time point; Indicates The cluster center of the drying mode is the historical drying sequence.

[0097] The adaptive PID parameters are obtained based on the similarity results. The PID parameters are obtained by weighted average of the optimal parameters of each drying mode, and the weight is determined by the similarity.

[0098] The calculation formula of the adaptive proportional gain parameter is: , the calculation formula of the adaptive integral gain parameter is , the calculation formula of adaptive differential gain is: .

[0099] in, , and They respectively represent the updated proportional gain parameter, updated integral gain parameter and updated differential gain parameter corresponding to the g-th parameter update time point in the current drying process; Indicates the drying sequence corresponding to the gth parameter update time point and the gth parameter update time point of the current drying process. The similarity of the cluster centers of the drying modes; , and Respectively represent The optimal proportional gain parameter, optimal integral gain parameter and optimal differential gain parameter of the g-th parameter update time point are obtained; M represents the number of drying modes. The PID parameters between the g-th parameter update time point and the g+1-th parameter update time point are adjusted by using the updated proportional gain parameter, updated integral gain parameter and updated differential gain parameter, that is, the PID parameters between the g-th parameter update time point and the g+1-th parameter update time point are replaced by the updated proportional gain parameter, updated integral gain parameter and updated differential gain parameter corresponding to the g-th parameter update time point.

[0100] Through the above steps, the similarity can be effectively used to calculate the adaptive PID parameters, so as to realize adaptive control. At the parameter update time point, the updated proportional gain parameters, updated integral gain parameters and updated differential gain parameters corresponding to the parameter update time point are calculated, and the subsequent PID parameters are adaptively adjusted. The adaptive proportional, integral and differential gain parameters are used for control, thereby improving the response speed and stability of the system and completing the intelligent control of temperature.

[0101] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0102] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0103] 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 spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A temperature intelligent control method for edible glue drying process, characterized in that: The method includes: Obtain a historical drying sequence, the historical drying sequence including the temperature and humidity of a drying process; calculate the temperature change rate and humidity change rate corresponding to each temperature and humidity in the historical drying sequence; calculate a multivariate distance metric for matching data point pairs in two historical drying sequences based on the temperature change rate and humidity change rate; Calculate the weight corresponding to the multivariate distance metric of the matching data point pair according to the temperature difference and humidity difference in the matching data point pair; calculate the distance between two historical drying sequences using the multivariate distance metric of the matching data point pair and its corresponding weight; Clustering is performed based on the distance between the historical drying sequences, and each category is recorded as a drying mode; and the optimal PID parameters of each drying mode are obtained according to the PID parameters corresponding to the historical drying sequences in the drying mode; Obtain the similarity between the drying sequence before each parameter update time point in the current drying process and the cluster center corresponding to each drying mode; adjust the PID parameters of the current drying process according to the similarity and the optimal PID parameters of each drying mode; The multivariate distance metric for matching pairs of data points in the two historical drying sequences is: ， in, represents a multivariate distance measure for pairs of matching data points, and They represent the data point of the historical drying sequence X at time i and the data point of the historical drying sequence Y at time j respectively; and Respectively represent the temperature and humidity of the historical drying sequence X at time i; and represent the temperature and humidity of the historical drying sequence Y at time j respectively; and They represent the relative rate of change of temperature and the relative rate of change of humidity respectively; The relative change rates of temperature and humidity are: ， ， in, and represents the temperature change rate and humidity change rate at time i in the historical drying sequence X; and They represent the temperature change rate and humidity change rate at time j in the historical drying sequence Y respectively; | | represents the absolute value symbol; The weights corresponding to the multivariate distance metric for the matching data point pairs are: ， in, Represents the weights corresponding to the multivariate distance measure for matching pairs of data points; and Respectively represent the temperature and humidity of the historical drying sequence X at time i; and They represent the temperature and humidity of the historical drying sequence Y at time j respectively; k represents the adjustment parameter of the sensitivity function; σ represents the adjustment parameter of the time distance; Represents a series of index pairs in the DTW algorithm The alignment paths of the components; represents the exponential function with the natural constant e as the base; N represents the number of all matching data point pairs in the DTW metric; The distance between the two historical drying sequences is: ， in, Represents the distance between historical drying sequence X and historical drying sequence Y; Represents the weights corresponding to the multivariate distance measure for matching pairs of data points; represents a multivariate distance measure for pairs of matching data points, and They represent the data point of the historical drying sequence X at time i and the data point of the historical drying sequence Y at time j respectively; Represents a series of index pairs in the DTW algorithm The alignment paths of the components; Represents the set of all alignment paths The obtaining of the similarity between the drying sequence before each parameter update time point in the current drying process and the cluster center corresponding to each drying mode includes: In the current drying process, parameter update time points are set at preset intervals; the temperature and humidity data before each parameter update time point are obtained to form a drying sequence; the similarity calculation formula is: ， in, Indicates the drying sequence corresponding to the gth parameter update time point and the gth parameter update time point of the current drying process. The similarity of the cluster centers of the drying modes; is the drying sequence corresponding to the current drying process from the beginning to the g-th parameter update time point; Indicates The cluster centers of the drying modes; exp( ) represents the exponential function with the natural constant as the base, and DTW( ) represents the dynamic time warping algorithm; The step of adjusting the PID parameters of the current drying process according to the similarity and the optimal PID parameters of each drying mode includes: Calculate the updated proportional gain parameter, updated integral gain parameter and updated differential gain parameter corresponding to the g-th parameter update time point in the current drying process; the calculation formulas are , and ; in, , and They respectively represent the updated proportional gain parameter, updated integral gain parameter and updated differential gain parameter corresponding to the g-th parameter update time point in the current drying process; Indicates the drying sequence corresponding to the gth parameter update time point and the gth parameter update time point of the current drying process. The similarity of the cluster centers of the drying modes; , and Respectively represent The optimal proportional gain parameter, optimal integral gain parameter and optimal differential gain parameter of each drying mode; M represents the number of drying modes; The PID parameters between the g-th parameter update time point and the g+1-th parameter update time point are adjusted by updating the proportional gain parameter, updating the integral gain parameter and updating the differential gain parameter.

2. The temperature intelligent control method for edible glue drying process according to claim 1 is characterized in that: The obtaining of the historical drying sequence, where the historical drying sequence includes the temperature and humidity of a drying process, includes: The temperature data and humidity data of the drying process are collected by using a temperature sensor and a humidity sensor respectively, and the sampling interval is a preset sampling interval; the temperature data and humidity data of a drying process constitute a historical drying sequence corresponding to the drying process.

3. The temperature intelligent control method for edible glue drying process according to claim 1 is characterized in that: The calculating of the temperature change rate and humidity change rate corresponding to each temperature and humidity in the historical drying sequence respectively includes: Obtain the difference between the temperature at a later moment and the temperature at a previous moment in the historical drying sequence, and the ratio of the difference to the time interval between the two moments is the temperature change rate corresponding to the temperature at the later moment; The difference between the humidity at a later moment and the humidity at a previous moment in the historical drying sequence is obtained, and the ratio of the difference to the time interval between the two moments is the humidity change rate corresponding to the humidity at the later moment.

4. The method for intelligent temperature control in the drying process of edible glue according to claim 1, characterized in that: The step of obtaining the optimal PID parameters of each drying mode according to the PID parameters corresponding to the historical drying sequences in the drying mode includes: The optimal PID parameters of each drying mode include the optimal proportional gain parameter, the optimal integral gain parameter and the optimal differential gain parameter; The average of the proportional gain parameters in the PID parameters corresponding to the preset number of historical drying sequences with the highest drying effect scores in each drying mode is the optimal proportional gain parameter of the drying mode; The average of the integral gain parameters in the PID parameters corresponding to the preset number of historical drying sequences with the highest drying effect scores in each drying mode is the optimal integral gain parameter of the drying mode; The average of the differential gain parameters in the PID parameters corresponding to a preset number of historical drying sequences with the highest drying effect scores in each drying mode is the optimal differential gain parameter of the drying mode.

5. The temperature intelligent control method for edible glue drying process according to claim 1 is characterized in that: After obtaining the historical drying sequence, it also includes: Experts are used to score the drying effect of each historical drying sequence to obtain the drying effect score of each historical drying sequence.

Images