Intelligent monitoring and regulating method for fried food processing process
By performing STL decomposition and AMPD lookup on oil temperature data, combined with the ARIMA model, the peak oil temperature point is identified and its weights are adjusted, solving the problem of untimely oil temperature adjustment in PID control. This enables accurate prediction and timely adjustment of oil temperature, improving the quality and safety of fried foods.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-20
- Publication Date
- 2026-03-13
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In existing PID-based fried food processing, the oil temperature adjustment is not timely enough, resulting in large oil temperature fluctuations, which affects food quality and safety.
The STL time series decomposition method is used to analyze historical oil temperature data, identify peak points in the periodic and residual terms, determine the peak points of feeding temperature changes through the AMPD peak finding algorithm, and determine the weights based on their distribution regularity and characteristics. Combined with the ARIMA model, oil temperature is predicted to improve the timeliness of adjustment.
By accurately predicting the trend of oil temperature changes, timely adjustment of oil temperature is achieved, which improves the control precision and stability in the processing of fried foods, and ensures food quality and safety.
Smart Images
Figure CN121657773A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and in particular to an intelligent monitoring and control method for the processing of fried foods. Background Technology
[0002] Monitoring and controlling oil temperature during the processing of fried foods is a key factor affecting food quality and safety. Properly controlling the oil temperature can prevent food quality and safety risks caused by abnormal oil temperatures. Current technologies typically employ PID control algorithms to monitor and control oil temperature. When a drop in oil temperature is detected, the PID control algorithm increases the heating power to bring the oil temperature back to the temperature required for the standard processing procedure.
[0003] However, during processing, the addition of ingredients can cause the oil temperature to drop instantly, especially in large-scale food processing where the temperature drop is even greater. Due to its working principle, the PID system can only start adjusting after detecting changes in oil temperature. In such cases, the adjustment timeliness is poor, and the huge fluctuations in oil temperature may affect the quality of fried food.
[0004] Therefore, current PID-based oil temperature control methods in fried food processing suffer from the technical problem of insufficient timeliness in adjusting oil temperature. Summary of the Invention
[0005] In view of this, the present invention provides an intelligent monitoring and control method for the processing of fried foods, in order to solve the technical problem that the current PID-based fried food processing is not timely in adjusting the oil temperature.
[0006] The present invention provides an intelligent monitoring and control method for the processing of fried foods, comprising: In the process of frying food, the oil temperature is controlled by PID control method, and historical oil temperature data of the first preset time before the current moment is collected. The historical oil temperature data is decomposed by STL time series to obtain periodic terms and residual terms. Determine the peak points in the periodic term and the residual term, take the peak points in the periodic term as significant periodic peak points, determine the feeding temperature change peak points caused by feeding based on the time difference between each peak point in the residual term and its adjacent peak points and the oil temperature difference, and determine the second prediction weight of the feeding temperature change peak points according to the degree of distribution regularity of the feeding temperature change peak points in the historical oil temperature data. The historical oil temperature data is used to fit and predict the predicted oil temperature data for a second preset time after the current moment. During the fitting and prediction process, the fitting coefficient of the significant periodic peak point in the historical oil temperature data is adjusted according to the preset first prediction weight, and the fitting coefficient of the feeding temperature change peak point in the historical oil temperature data is adjusted according to the second prediction weight. The first prediction weight and the second prediction weight are both greater than 1, and the first prediction weight is greater than the second prediction weight. Based on the predicted oil temperature data, perform PID control of the oil temperature after the current moment.
[0007] Furthermore, determining the peak point in the periodic term and the residual term includes: The AMPD peak finding algorithm is used to determine the peak points in the periodic term and the residual term.
[0008] Furthermore, determining the peak point of the temperature change caused by feeding includes: The oil temperature fluctuation amplitude term is obtained by summing the absolute values of the oil temperature difference between any peak point in the residual term and the two adjacent peak points on the left and right, and normalizing them. The oil temperature fluctuation duration term is obtained by summing the time intervals between any peak point in the residual term and the two adjacent peak points on the left and right, and normalizing them. The sum of the oil temperature fluctuation amplitude term and the oil temperature fluctuation duration term is taken as the feeding fluctuation probability of any peak point in the residual term. The peak point in the residual term where the feeding fluctuation probability is greater than the preset feeding fluctuation probability threshold is taken as the feeding temperature change peak point caused by feeding.
[0009] Furthermore, it also includes: The first characteristic of the current feeding temperature change peak point is determined based on the difference between the oil temperature value at the current feeding temperature change peak point and the average oil temperature value of all data points in the residual term; The second characteristic of the current feeding temperature peak point is determined based on the degree of oil temperature change between any two adjacent feeding temperature peak points within a preset range from the current feeding temperature peak point. Based on the first feature and the second feature, all the peak points of the feeding temperature change are clustered, and the second prediction weight of the peak points of the feeding temperature change under the corresponding cluster is determined according to the degree of distribution regularity of the peak points of the feeding temperature change under each cluster in the historical oil temperature data.
[0010] Furthermore, the first feature for determining the current peak temperature change point of the feed includes: The absolute value of the difference between the oil temperature at the current peak temperature of the feeding process and the average oil temperature of all data points in the residual term is taken as the first characteristic of the current peak temperature of the feeding process.
[0011] Furthermore, the second feature for determining the current peak temperature change point of the feed includes: After selecting feeding temperature change peak points within the preset range on the left and right sides of the current feeding temperature change peak point, the selected feeding temperature change peak points and the current feeding temperature change peak point constitute the feeding temperature change peak point set corresponding to the current feeding temperature change peak point. Calculate the absolute value of the difference in oil temperature between any two adjacent peak values of the current feeding temperature change, and sum the absolute values of the differences in all oil temperature values as the second characteristic of the current feeding temperature change peak value.
[0012] Furthermore, the second prediction weight for determining the peak point of the feeding temperature change includes: Calculate the interval between any two adjacent peak temperature points of the feed in the historical oil temperature data, calculate the absolute value of the difference between any interval and the mean interval and record it as the irregularity of the corresponding interval, and record the mean of the irregularity corresponding to all intervals as the overall irregularity. The second prediction weight is determined based on the overall irregularity, and the second prediction weight is inversely proportional to the overall irregularity.
[0013] Furthermore, the step of performing PID control of the oil temperature after the current moment based on the predicted oil temperature data includes: Taking any time after the current time as the target time, and based on the predicted oil temperature at a time after the target time with a third preset time interval from the target time, PID control of the oil temperature at the target time is completed.
[0014] The advantages of this invention compared to the prior art are: This invention performs time-series decomposition on historical oil temperature data controlled by PID during the processing of fried foods. Peak points in the decomposed periodic and residual terms are selected as periodic change characterization points. These peak points directly represent oil temperature changes caused by normal process variations and material input changes. In the residual term, the differences in data change characteristics of peak points under actual material input and environmental influences are used to determine the oil temperature change data points truly caused by material input. Then, based on the characterization ability of each selected oil temperature change data point, the contribution of each data point in the fitting process is adjusted when using historical oil temperature data to predict future oil temperatures, resulting in a more accurate predicted oil temperature. Finally, based on the accurately predicted oil temperature, the PID control of the current oil temperature is completed according to the predicted future oil temperature change trend, improving the timeliness of oil temperature adjustment in PID-controlled fried food processing. Attached Figure Description
[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 This is a flowchart illustrating an intelligent monitoring and control method for the processing of fried food provided in Embodiment 1 of the present invention. Detailed Implementation
[0017] The overall concept of this invention is as follows: The historical oil temperature data under PID control prior to the current moment is decomposed using STL time series analysis. The peak points in the decomposed periodic terms are recorded as data points with significant periodic feeding variation characteristics and assigned a first-class weight. Then, the peak points in the residual terms, which may be caused by feeding or environmental influences, are distinguished to screen out the feeding temperature change peak points caused by feeding. The predictability of each feeding temperature change peak point is further quantified to determine its corresponding second-class weight. Based on the first-class and second-class weights, the fitting prediction coefficients of the corresponding data points in the historical oil temperature data are adjusted to obtain more accurate oil temperature prediction data. Based on the oil temperature prediction value in the short future, the PID control of the oil temperature at the current moment is completed, improving the timeliness of oil temperature regulation in the fried food processing process.
[0018] To further illustrate the technical solution of the present invention, specific embodiments are described below.
[0019] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of the invention include a particular feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. Furthermore, a particular feature, structure, or characteristic in one or more embodiments may be combined in any suitable form, and the terms "comprising," "including," "having," and variations thereof mean "including, but not limited to," unless otherwise specifically emphasized.
[0020] It should be understood that the sequence number of each step in the following embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
[0021] Method Implementation Examples: See Figure 1 This is a flowchart illustrating an intelligent monitoring and control method for the processing of fried food provided in Embodiment 1 of the present invention. Figure 1 As shown, the monitoring and control method may include the following steps: S101, In the process of frying food processing, the oil temperature is controlled by PID control method, and historical oil temperature data of a first preset time period before the current time is collected. The historical oil temperature data is decomposed by STL time series to obtain periodic terms and residual terms.
[0022] In the process of frying food, an oil temperature sensor is used to monitor the temperature inside the oil pan, and a PID control system is used to adjust the oil temperature in real time based on the monitored temperature, keeping the oil temperature stable near the calibrated frying temperature. The preferred oil temperature monitoring frequency is once per second, and the monitoring data is transmitted to a data center.
[0023] Considering the need to minimize computation and the low reference value of early historical data in the prediction process, this embodiment preferably extracts oil temperature monitoring data from the last hour before the current moment as historical oil temperature data. The historical oil temperature data is projected onto a two-dimensional plane coordinate system, with the horizontal axis representing time and the vertical axis representing oil temperature. This allows for the determination of each oil temperature data point corresponding to the monitoring frequency.
[0024] Because frying typically requires repeated frying at different oil temperatures, the oil temperature during the frying process itself exhibits a periodic variation pattern. Furthermore, in certain cases, such as the processing of a particular product where the processing steps are clearly defined and fixed, the frequency of ingredient addition also shows a clear regularity, resulting in periodicity in the oil temperature monitoring data. Therefore, these data segments with obvious periodicity can be obtained through time-series decomposition. In this embodiment, STL time-series decomposition is preferably performed on historical oil temperature data to obtain periodic terms, trend terms, and residual terms.
[0025] S102, determine the peak points in the periodic term and the residual term, take the peak points in the periodic term as significant periodic peak points, determine the feeding temperature change peak points caused by feeding based on the time difference between each peak point in the residual term and its adjacent peak points and the oil temperature difference, and determine the second prediction weight of the feeding temperature change peak points according to the degree of distribution regularity of the feeding temperature change peak points in the historical oil temperature data.
[0026] The periodic terms obtained from STL time-series decomposition can characterize significant periodic fluctuations in historical oil temperature data. Within these periodic terms, the AMPD peak-finding algorithm is preferably used to obtain each peak and trough point, i.e., to determine each peak point. These peak points exhibit significant periodicity, representing periodic changes in oil temperature caused by process requirements (such as different oil temperatures required for primary pressing and secondary frying) or significant periodic changes in oil temperature caused by feeding. These peak points in the periodic terms, or the data points corresponding to each peak point in the periodic terms in the historical oil temperature data, are considered significant periodic peak points and uniformly assigned a first prediction weight for predicting subsequent oil temperature changes.
[0027] Meanwhile, considering that the feeding situation in the processing of fried foods does not necessarily conform to a single periodicity, but often involves complex feeding situations, the periodic term obtained after STL time series decomposition will not include all temperature change data points caused by feeding. Therefore, this embodiment continues to analyze and determine the temperature change data points caused by feeding in the residual term, and obtain whether these temperature change data points caused by feeding have a regularity and the magnitude of the regularity. If a strong regularity exists, a second prediction weight can be assigned based on the magnitude of its regularity, which can also be used for subsequent oil temperature change prediction.
[0028] Specifically, the first step is to analyze and determine the temperature change data points caused by material feeding in the residual terms.
[0029] During the processing of fried foods, the oil temperature changes contained in the residual data are not only caused by the addition of ingredients, but also by environmental interference, external humidity and temperature conduction, etc. These non-ingredient oil temperature changes can affect the accuracy of subsequent oil temperature prediction. Therefore, it is necessary to first distinguish and determine whether these oil temperature change data points are caused by the addition of ingredients.
[0030] Environmental disturbances, external humidity and temperature conduction, etc., cause relatively small fluctuations in oil temperature. Due to the existence of the PID control algorithm, after detecting changes in oil temperature caused by external environmental disturbances, the system will immediately adjust the boiler heating power to restore the oil temperature to the normal value. Therefore, the fluctuation amplitude of this type of change is relatively small, and the appearance and end of the fluctuation are also relatively quick, that is, the duration is short. However, feeding will cause a large change in oil temperature. Therefore, the fluctuation amplitude of oil temperature change data caused by feeding will be relatively larger. Moreover, the PID control algorithm has a time delay in adjusting for large oil temperature changes and requires a certain amount of time to recover. The appearance and end of the corresponding fluctuation has a certain duration, that is, the duration is longer.
[0031] Meanwhile, considering that the periodicity is most strongly represented by peak points in the residual term, the AMPD peak finding algorithm is also used to obtain each peak and trough point in the residual term, i.e., to determine each peak point. Then, combining the distinguishing features between the oil temperature change caused by feeding and the oil temperature change caused by interference, the peak points of the feeding temperature change in the residual term are determined, including: The oil temperature fluctuation amplitude term is obtained by summing the absolute values of the oil temperature difference between any peak point in the residual term and the two adjacent peak points on the left and right, and normalizing them. The oil temperature fluctuation duration term is obtained by summing the time intervals between any peak point in the residual term and the two adjacent peak points on the left and right, and normalizing them. The sum of the oil temperature fluctuation amplitude term and the oil temperature fluctuation duration term is taken as the feeding fluctuation probability of any peak point in the residual term. The peak point in the residual term where the feeding fluctuation probability is greater than the preset feeding fluctuation probability threshold is taken as the feeding temperature change peak point caused by feeding.
[0032] The probability of fluctuation in material input is as follows:
[0033] in, Let q be the probability of material fluctuation at the q-th peak point in the residual term. This represents the sum of the longitudinal distances from the q-th peak point to its left and right adjacent peak points. The larger this sum of longitudinal distances, the greater the sum of the absolute values of the oil temperature differences between the q-th peak point and its left and right adjacent peak points, indicating a larger fluctuation range. This better matches the characteristic of a significant change in oil temperature during material feeding, and suggests a higher probability of successful material feeding. This represents the sum of the lateral distances from the q-th peak point to its left and right adjacent peak points. The larger this sum of lateral distances, the larger the sum of the time intervals between the q-th peak point and its two adjacent peak points, indicating a longer duration of fluctuation. This better matches the characteristic of a longer callback time in the PID system after feeding, and a higher probability of feeding. To achieve linear normalization of the maximum and minimum values, the calculated values for this part of the formula are normalized based on all peak points to ensure that the two parts of the formula have the same magnitude.
[0034] Based on the calculated feeding fluctuation probability of each peak point in the residual term, peak points with feeding characteristics and peak points without feeding characteristics are divided into two ends of the feeding fluctuation probability range. Here, the preferred threshold value of 1 in the feeding fluctuation probability range is used as the preset feeding fluctuation probability threshold. When the feeding fluctuation probability of any peak point in the residual term... At that time, the peak point is considered to be the data point of temperature change fluctuation caused by feeding, that is, the peak point of temperature change caused by feeding.
[0035] After determining the peak points of each feed temperature change in the residual term, the second prediction weight of the peak points of each feed temperature change can be determined based on the degree of distribution regularity of each peak point of feed temperature change in historical oil temperature data.
[0036] However, considering that in the process of frying food processing, even for the same processed product, the raw materials used in the production process may include a variety of ingredients, and the oil temperature will change differently due to the different sizes and amounts of each raw material.
[0037] Therefore, in a preferred embodiment, after determining the peak points of each feed temperature change in the residual term, the method further includes: The first characteristic of the current feeding temperature change peak point is determined based on the difference between the oil temperature value at the current feeding temperature change peak point and the average oil temperature value of all data points in the residual term; The second characteristic of the current feeding temperature peak point is determined based on the degree of oil temperature change between any two adjacent feeding temperature peak points within a preset range from the current feeding temperature peak point. Based on the first feature and the second feature, all the peak points of the feeding temperature change are clustered, and the second prediction weight of the peak points of the feeding temperature change under the corresponding cluster is determined according to the degree of distribution regularity of the peak points of the feeding temperature change under each cluster in the historical oil temperature data.
[0038] The reasons for constructing the first and second features mentioned above are twofold: On the one hand, different feeds produce different oil temperature changes, which means that different degrees of outliers will be generated in the residual term. Therefore, the degree of outlier can be used as a feature for similarity recognition in the clustering process. On the other hand, different feed materials react differently when in contact with hot oil. Some feed materials are small and fragmented, but dense, with a large contact area with hot oil. Therefore, the oil temperature conduction rate is faster when the feed materials come into contact with the hot oil, and the oil temperature changes more rapidly. This results in greater differences among other residual points near the residual point in the corresponding residual term. On the other hand, those relatively large and intact feed materials have a relatively small contact area with hot oil, and the feed materials need to be continuously heated internally. Therefore, the oil temperature conduction rate is slower when the feed materials come into contact with the hot oil, and the oil temperature changes more slowly. This results in smaller differences among other residual points near the residual point in the corresponding residual term. Therefore, the rate of change of oil temperature can be used as another feature for similarity recognition in the clustering process.
[0039] As a further preferred method, the first feature is obtained as follows:
[0040] Wherein, represents the first characteristic of the p-th peak temperature change point during material feeding. This represents the oil temperature value at the p-th peak temperature point of the feeding process. This represents the average oil temperature value at all peak temperature points during material feeding. This indicates taking the absolute value. The difference between the oil temperature value at the p-th feeding temperature change peak point and the average oil temperature value at all other feeding temperature change peak points is considered. The larger the difference, the greater the outlier the p-th feeding temperature change peak point is.
[0041] The second feature is obtained as follows:
[0042] in, This is the second characteristic of the p-th peak temperature change point during material feeding. This is the number of the leftmost adjacent peak values of the temperature change of the p-th feeding point, which is also the preset range from the left side of the current p-th feeding peak value. This is the number of the number of the nearest temperature change peak points to the right of the p-th feeding temperature change peak point, which is also the preset range between the right side of the current p-th feeding temperature change peak point and the current p-th feeding temperature change peak point. In this further preferred embodiment, and The value of is set to 3, which is used to determine the set of feeding temperature change peak points corresponding to the p-th feeding temperature change peak point. and These represent the oil temperature values on the left and right sides of the set of peak feeding temperature changes corresponding to the p-th peak feeding temperature change point, respectively, at the i-th or j-th peak feeding temperature change point. and These represent the difference in oil temperature between any two adjacent peak points in the set of peak temperature changes during material feeding. The larger the sum of these differences, the faster the oil temperature changes upon material contact.
[0043] The first and second features of all feeding temperature change data points are obtained. These two features are used as two feature dimensions to construct a sample space. K-means clustering is then performed on all feeding temperature change data points, where the number of target clusters, k, can be determined using the elbow method. The clustering results will group feeding data points with similar features into one cluster, resulting in multiple clusters of feeding data points, each cluster representing a similar type of feeding data point. The similar feeding data points obtained thus exhibit high similarity in the change of oil temperature caused by feeding. Therefore, when determining the second prediction weight of the peak feeding temperature change point, it can be determined according to the second prediction weight of the peak feeding temperature change point based on the clusters.
[0044] In determining the second prediction weight, the more regular the distribution of all peak temperature points of the feed input in the historical oil temperature data, the more regular the feed input behavior of all peak temperature points participating in the calculation, and the stronger its predictability. Therefore, the corresponding second prediction weight should be larger. Thus, the second prediction weight of the peak temperature points of the feed input can be determined, including: Calculate the interval between any two adjacent peak temperature points of the feed in the historical oil temperature data, calculate the absolute value of the difference between any interval and the mean interval and record it as the irregularity of the corresponding interval, and record the mean of the irregularity corresponding to all intervals as the overall irregularity. The second prediction weight is determined based on the overall irregularity, and the second prediction weight is inversely proportional to the overall irregularity.
[0045] Corresponding to the above preferred embodiment, all peak points of temperature change in the residual term were clustered. Here, the formula for determining the second prediction weight with a single cluster as the calculation range is further given:
[0046] in, The second prediction weight is used for each peak temperature change point of the feed input within the c-th cluster. The number of peak temperature change points during feeding within the c-th cluster. This indicates the number of intervals between these peak temperature change points during material feeding. Let be the interval between the i-th peak temperature change point during material feeding and the next peak temperature change point during material feeding, as defined in the historical oil temperature data. It is the average time interval between any two adjacent peak values of temperature change during feeding within the c-th cluster. This represents the average difference between the interval duration and the mean of the interval durations among all the peak temperature change points within the c-th cluster. The smaller this average difference, the more consistent the time intervals of the peak temperature change points within the c-th cluster, indicating stronger predictability and a larger second prediction weight. Therefore, an inverse proportional normalization function is used. .
[0047] S103, using the historical oil temperature data to fit and predict the predicted oil temperature data for a second preset time period after the current moment, and adjusting the fitting coefficient of the significant periodic peak point corresponding to the data point in the historical oil temperature data according to the preset first prediction weight, and adjusting the fitting coefficient of the feeding temperature change peak point corresponding to the data point in the historical oil temperature data according to the second prediction weight, wherein the first prediction weight and the second prediction weight are both greater than 1 and the first prediction weight is greater than the second prediction weight.
[0048] After determining the first prediction weight of each significant periodic peak point in the periodic term and the second prediction weight of each feeding temperature change peak point in the residual term, the fitting coefficients of relevant data points in the historical oil temperature data used for fitting prediction can be adjusted according to the prediction weights during the fitting prediction process of the prediction model, so as to obtain the predicted oil temperature data for the second preset time period after the current moment.
[0049] Specifically, since the periodic terms obtained from the aforementioned time series decomposition have more typical periodic characteristics than the residual terms, the first prediction weight is set to be greater than the second prediction weight. Furthermore, to increase the contribution of the significant periodic peak points and the peak points of the feeding temperature change obtained above in the fitting prediction process, both the first and second prediction weights should be greater than 1. In this embodiment, the ARIMA model is preferably used as the prediction model for fitting prediction.
[0050] Based on the settings of the first prediction weight and the second prediction weight, during the fitting prediction process, the fitting coefficient of the significant periodic peak point corresponding to the data point in the historical oil temperature data is adjusted according to the preset first prediction weight, and the fitting coefficient of the feeding temperature change peak point corresponding to the data point in the historical oil temperature data is adjusted according to the second prediction weight. A higher prediction weight indicates that the corresponding peak point belongs to a more regular oil temperature change data point caused by feeding or process changes. Therefore, the oil temperature change prediction obtained based on this data point is more accurate, and the influence of this data point in the prediction fitting should be higher.
[0051] Preferably, the adjusted fitting coefficients of the significant periodic peak points to the corresponding data points in the historical oil temperature data, and the adjusted fitting coefficients of the peak points of feed temperature changes to the corresponding data points in the historical oil temperature data, are as follows:
[0052]
[0053] in, This represents the adjusted fit coefficient of the data point corresponding to the z-th significant period peak in the historical oil temperature data. represents the original fitting coefficient of the data point corresponding to the z-th significant period peak point in the historical oil temperature data, and 2 represents the adjustment factor of the fitting coefficient of the data point corresponding to the significant period peak point selected according to the second prediction weight, which is determined according to the value range of the second prediction weight. This means doubling the original fitting coefficients to increase their proportion in the fitting weights, making the predictions more accurate for significant periodic changes in oil temperature. This represents the adjusted fitting coefficient of the data point corresponding to the t-th feed temperature change peak point in the historical oil temperature data. This represents the second prediction weight corresponding to the t-th peak temperature change point during material feeding. This represents the original fitting coefficient of the data point corresponding to the t-th feed temperature change peak point in the historical oil temperature data. This indicates that, based on the original fitting coefficients, the higher the second prediction weight of the peak temperature change point of the feed, the greater its influence on the fitting.
[0054] The historical oil temperature data after adjusting the fitting coefficient is used to perform ARIMA fitting and prediction to obtain the predicted oil temperature data for the second preset time after the current moment.
[0055] S104, Based on the predicted oil temperature data, complete the PID control of the oil temperature after the current moment.
[0056] For a PID control system, after obtaining the predicted oil temperature data for a future period of time, the PID control system can select the predicted oil temperature at a time three preset time intervals after the target time to determine the oil temperature adjustment target at the target time, so as to achieve advance adjustment and improve the timeliness of adjustment.
[0057] As the predicted oil temperature data is continuously acquired iteratively, the target time can be any time after the current time. In this embodiment, the third preset duration is preferably five seconds, that is, the predicted oil temperature five seconds after the target time is selected for adjusting the oil temperature at the target time. The specific value of the third preset duration can be determined by comprehensively considering the requirements for the degree of oil temperature pre-adjustment and the requirements for oil temperature adjustment stability. The larger the value of the third preset duration, the greater the degree of oil temperature pre-adjustment and the higher the degree of pre-response to future oil temperature fluctuations. However, since this pre-response is based on the predicted oil temperature, there will inevitably be a deviation from the actual future oil temperature change. Therefore, the larger the value of the third preset duration, the lower the oil temperature adjustment stability.
[0058] Furthermore, the so-called selection of the predicted oil temperature at a time interval of three preset time after the target time to determine the oil temperature adjustment target at the target time can be either directly using the predicted oil temperature at a time interval of three preset time as the PID control target oil temperature at the target time, or it can be obtaining the PID control target oil temperature by weighting the actual monitored oil temperature at the target time and the predicted oil temperature at a time interval of three preset time with their respective weighting values and then summing them up to obtain the combined oil temperature data from both aspects.
[0059] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.
Claims
1. An intelligent monitoring and control method for the processing of fried foods, characterized in that, The method includes: In the process of frying food, the oil temperature is controlled by PID control method, and historical oil temperature data of the first preset time before the current moment is collected. The historical oil temperature data is decomposed by STL time series to obtain periodic terms and residual terms. Determine the peak points in the periodic term and the residual term, take the peak points in the periodic term as significant periodic peak points, determine the feeding temperature change peak points caused by feeding based on the time difference between each peak point in the residual term and its adjacent peak points and the oil temperature difference, and determine the second prediction weight of the feeding temperature change peak points according to the degree of distribution regularity of the feeding temperature change peak points in the historical oil temperature data. The historical oil temperature data is used to fit and predict the predicted oil temperature data for a second preset time after the current moment. During the fitting and prediction process, the fitting coefficient of the significant periodic peak point in the historical oil temperature data is adjusted according to the preset first prediction weight, and the fitting coefficient of the feeding temperature change peak point in the historical oil temperature data is adjusted according to the second prediction weight. The first prediction weight and the second prediction weight are both greater than 1, and the first prediction weight is greater than the second prediction weight. Based on the predicted oil temperature data, perform PID control of the oil temperature after the current moment.
2. The intelligent monitoring and control method for the fried food processing process according to claim 1, characterized in that, Determining the peak point in the periodic term and the residual term includes: The AMPD peak finding algorithm is used to determine the peak points in the periodic term and the residual term.
3. The intelligent monitoring and control method for the fried food processing process according to claim 1, characterized in that, Determining the peak point of temperature change caused by material feeding includes: The oil temperature fluctuation amplitude term is obtained by summing the absolute values of the oil temperature difference between any peak point in the residual term and the two adjacent peak points on the left and right, and normalizing them. The oil temperature fluctuation duration term is obtained by summing the time intervals between any peak point in the residual term and the two adjacent peak points on the left and right, and normalizing them. The sum of the oil temperature fluctuation amplitude term and the oil temperature fluctuation duration term is taken as the feeding fluctuation probability of any peak point in the residual term. The peak point in the residual term where the feeding fluctuation probability is greater than the preset feeding fluctuation probability threshold is taken as the feeding temperature change peak point caused by feeding.
4. The intelligent monitoring and control method for the fried food processing process according to claim 1 or 3, characterized in that, Also includes: The first characteristic of the current feeding temperature change peak point is determined based on the difference between the oil temperature value at the current feeding temperature change peak point and the average oil temperature value of all data points in the residual term; The second characteristic of the current feeding temperature peak point is determined based on the degree of oil temperature change between any two adjacent feeding temperature peak points within a preset range from the current feeding temperature peak point. Based on the first feature and the second feature, all the peak points of the feeding temperature change are clustered, and the second prediction weight of the peak points of the feeding temperature change under the corresponding cluster is determined according to the degree of distribution regularity of the peak points of the feeding temperature change under each cluster in the historical oil temperature data.
5. The intelligent monitoring and control method for the processing of fried foods according to claim 4, characterized in that, The first feature for determining the current peak temperature change point of the feed includes: The absolute value of the difference between the oil temperature at the current peak temperature of the feeding process and the average oil temperature of all data points in the residual term is taken as the first characteristic of the current peak temperature of the feeding process.
6. The intelligent monitoring and control method for the processing of fried food according to claim 4, characterized in that, The second feature for determining the current peak temperature change point of the feed includes: After selecting feeding temperature change peak points within the preset range on the left and right sides of the current feeding temperature change peak point, the selected feeding temperature change peak points and the current feeding temperature change peak point constitute the feeding temperature change peak point set corresponding to the current feeding temperature change peak point. Calculate the absolute value of the difference in oil temperature between any two adjacent peak values of the current feeding temperature change, and sum the absolute values of the differences in all oil temperature values as the second characteristic of the current feeding temperature change peak value.
7. The intelligent monitoring and control method for the fried food processing process according to claim 1, characterized in that, The second prediction weight for determining the peak point of the feeding temperature change includes: Calculate the interval between any two adjacent peak temperature points of the feed in the historical oil temperature data, calculate the absolute value of the difference between any interval and the mean interval and record it as the irregularity of the corresponding interval, and record the mean of the irregularity corresponding to all intervals as the overall irregularity. The second prediction weight is determined based on the overall irregularity, and the second prediction weight is inversely proportional to the overall irregularity.
8. The intelligent monitoring and control method for the processing of fried food according to claim 1, characterized in that, The step of performing PID control of the oil temperature after the current moment based on the predicted oil temperature data includes: Taking any time after the current time as the target time, and based on the predicted oil temperature at a time after the target time with a third preset time interval from the target time, PID control of the oil temperature at the target time is completed.