A Dynamic Measurement and Compensation Method for Microwave Heating Temperature Based on Multi-Source Data Fusion

Abstract

This invention discloses a dynamic temperature measurement and compensation method for microwave heating based on multi-source data fusion, belonging to the field of dynamic temperature field monitoring and compensation technology. The method includes the following steps: 1. Using different types of temperature sensors, measuring the dynamic temperature data of materials during the process of multiple parameter changes under microwave heating, and then calculating the first-order difference between adjacent time points; 2. Calculating the proportion of volatility and negative change; 3. Calculating the stability score of temperature measurement data from different sensors, and then calculating the normalized weight of each sensor to ensure that the sum of the weights of multiple sensors is 1; 4. Calculating the weighted fusion temperature as the final temperature; 5. Regression analysis to construct a temperature measurement compensation model corresponding to each temperature sensor. This invention proposes an adaptive weighted fusion algorithm for multi-source temperature sensors based on dynamic stability assessment, solving the problem of limited accuracy of single sensors in existing temperature measurement scenarios, and improving the accuracy and stability of temperature measurement.

Description

Technical Field

[0001] This invention belongs to the field of dynamic temperature field monitoring and compensation technology, specifically relating to a method for dynamic measurement and compensation of microwave heating temperature based on multi-source data fusion. Background Technology

[0002] With the rapid development of industrial automation and intelligent manufacturing technologies, temperature monitoring and control play a crucial role in many industrial applications, especially in temperature regulation during material processing, heat treatment, and heating processes. Precise temperature control is essential for ensuring product quality and production efficiency. However, traditional temperature measurement methods such as thermocouples and infrared sensors typically rely on a single sensor for data acquisition. While this method is simple and easy to implement, it has the following drawbacks in complex and dynamic temperature measurement environments:

[0003] (1) Insufficient accuracy in temperature measurement. Traditional single sensors are often susceptible to noise, interference, and fluctuations in sensor performance when faced with electromagnetic interference, high temperature changes, atmospheric environmental fluctuations, or differences in the properties of materials, leading to unstable measurement data. For example, in complex electromagnetic environments, traditional single temperature measurement methods have significant limitations in microwave fields; thermocouples are easily affected by electromagnetic interference, resulting in significant nonlinear errors in the high-temperature range; infrared temperature measurement is affected by water vapor scattering and changes in surface emissivity in the low-temperature range, resulting in large error fluctuations; although fiber optic temperature measurement has strong anti-electromagnetic interference capabilities, it is limited by the temperature resistance of the material. These limitations lead to problems such as accuracy decay, response lag, and insufficient spatial resolution in temperature field monitoring data, making it difficult for the sensor to accurately reflect the approximate true temperature of the material during the temperature measurement process.

[0004] (2) Inconsistent response exists during temperature measurement. Because sensors may exhibit different response characteristics under different operating conditions, it cannot be guaranteed that all sensors will maintain the same accuracy throughout the entire measurement range. Different models of sensors have different response capabilities under electromagnetic interference and high temperatures, which may lead to large deviations in the measurement results of different sensors, making it impossible to accurately reflect temperature changes.

[0005] In summary, existing single-method temperature measurement techniques suffer from insufficient accuracy and are susceptible to variations in the response characteristics of different temperature sensors, leading to significant deviations in real-time temperature measurements. Furthermore, they are ill-suited for effectively monitoring dynamic temperature fields in complex electromagnetic and high-temperature environments. In addition, the coupling mechanism between process parameters and temperature measurement errors in microwave fields is not fully understood, resulting in insufficient adaptability of existing compensation methods. Moreover, most microwave heating equipment commonly used in industrial settings only possesses infrared temperature measurement capabilities. Summary of the Invention

[0006] To address the problem that existing single temperature measurement methods are insufficient to effectively meet the dynamic temperature field monitoring needs under complex electromagnetic and high-temperature environments, this invention provides a microwave heating temperature dynamic measurement and compensation method based on multi-source data fusion.

[0007] The technical solution of the present invention is as follows:

[0008] A method for dynamic measurement and compensation of microwave heating temperature based on multi-source data fusion, characterized by comprising the following steps:

[0009] Step 1: Using different types of temperature sensors, track and measure the dynamic temperature data of the material in real time during the microwave heating process of various key parameters. Then, calculate the first-order difference between adjacent time points based on the temperature data from each sensor. This refers to the magnitude of temperature change during the heating process; a positive number indicates that the temperature has increased compared to the previous measurement, while a negative number indicates that the temperature has decreased compared to the previous measurement.

[0010] ,

[0011] In the formula, This represents the temperature measurement value of the k-th sensor at time point i. This indicates the magnitude of temperature change during the heating process, where k represents the sensor type and i represents the time index.

[0012] Step 2: Calculate the fluctuation based on the temperature after first-order difference analysis from different sensors. and the proportion of negative changes :

[0013] ,

[0014] ,

[0015] In the formula, This indicates the magnitude of temperature fluctuation in the data measured by each temperature sensor, in μ. k Indicates the data passed through the k-th sensor. The calculated mean temperature change This represents the proportion of negative changes, i.e., the frequency of temperature decreases during the heating process; N represents the total number of changes.

[0016] Step 3: Calculate the stability score of temperature measurement data from different sensors based on the proportion of volatility and negative change. Then, the normalized weights of each sensor are calculated to ensure that the sum of the weights of all sensors is 1.

[0017] ,

[0018] ,

[0019] In the formula, This represents the score calculated based on the stability of temperature measurement data from different sensors. This indicates the normalized weighting of temperature measurements from each temperature sensor.

[0020] Step 4: Calculate the weighted fusion temperature as the final temperature based on the normalized weights of each temperature sensor and its corresponding temperature value.

[0021] ,

[0022] In the formula, This represents the approximate true temperature data calculated based on temperature measurement data from various temperature sensors and normalized weights.

[0023] Step 5: Further, using the temperature obtained from each temperature sensor as the independent variable and the weighted fused temperature as the dependent variable, a regression analysis is performed using a regression algorithm to construct a temperature measurement compensation model for each temperature sensor. The temperature measurement compensation model is used to accurately correct the original temperature measurement value of the sensor.

[0024] Preferably, the temperature sensor includes a thermocouple sensor, an infrared sensor, and a fiber optic sensor.

[0025] Preferably, in step one, the parameters include sample heating temperature, sample placement position, initial sample moisture content, and input power density.

[0026] Preferably, in step five, the regression analysis includes linear regression, quadratic polynomial regression, or trigonometric polynomial regression.

[0027] Further preferred, multiple sets of temperature measurement data under different parameter conditions are randomly divided into training and validation sets in a 7:3 ratio. The root mean square error (RMSEC) of the training set and the coefficient of determination (CDO) of the training set are used. Root mean square error of validation set (RMSEP) and coefficient of determination of validation set Evaluate the model.

[0028] Preferably, the method also includes step six: based on the actual sensor conditions of the microwave heating equipment, further embed the compensation model corresponding to each temperature sensor into the PLC program control system, and correct the original measurement value of the sensor by calling the compensation model in real time, and output the corrected temperature data to the equipment control module to achieve closed-loop precise control of microwave heating temperature.

[0029] Preferably, in step one, the interval between adjacent time points is a time length selected between 1 and 5 seconds. That is, the data acquisition frequency is to read a temperature value at a selected time interval every 1-5 seconds. For example, if 1 second is selected as the data acquisition frequency, and temperature data is read once at the 1st, 2nd, and 3rd seconds, then the 1st and 2nd seconds, and the 2nd and 3rd seconds are adjacent time points. More preferably, it is 1 second.

[0030] The beneficial effects of this invention are as follows:

[0031] This invention proposes an adaptive weighted fusion algorithm for multi-source temperature sensors based on dynamic stability assessment, addressing the limitation of single-sensor accuracy in existing temperature measurement scenarios and improving the accuracy and stability of temperature measurements. This method constructs a dual-index weight allocation model by comprehensively considering the volatility of the measurement sequence and the anomaly rate of negative temperature changes, achieving optimal fusion of data from different sensors. The resulting weighted temperature is more representative. Furthermore, the weighted temperature is used as an approximate true temperature, and regression analysis is performed with the temperatures measured by individual sensors, enabling the prediction of the approximate true temperature of materials using a single sensor.

[0032] The advantages of this invention are that, in order to address the shortcomings of traditional single sensors in complex temperature measurement scenarios, it adopts dynamic stability assessment and adaptive weighted fusion algorithm, thereby overcoming problems such as measurement error and temperature fluctuation. By weighted fusion of multi-source sensor data, a compensation model is obtained, and the output is the corrected sensor temperature value, which improves the accuracy and stability of temperature measurement. Furthermore, the temperature prediction effect is optimized through regression analysis, making it widely applicable to industrial applications that require high-precision temperature monitoring.

[0033] This invention is applicable to fields such as material heating, intelligent manufacturing, and industrial temperature control. It can provide more accurate and stable temperature measurement and prediction, especially under conditions of large temperature variations and complex environments. Attached Figure Description

[0034] Figure 1 The temperature measured by different sensors changes with heating time when the initial moisture content of the sample is 15%.

[0035] Figure 2 The change of temperature measured by different sensors with heating time when the sample is heated to 200℃;

[0036] Figure 3 The temperature measured by different sensors changes with heating time when the sample is placed at 0 cm.

[0037] Figure 4 The temperature change of different sensors with heating time when the sample input power density is 40 W / g;

[0038] Figure 5 Scatter plots of uncompensated temperature, compensated temperature, and approximate true temperature measured by the infrared sensor are shown, where (a) is the scatter plot of uncompensated temperature and approximate true temperature measured by the infrared sensor, and (b) is the scatter plot of compensated temperature and approximate true temperature measured by the infrared sensor. Scatter plot;

[0039] Figure 6 Scatter plots of uncompensated temperature, compensated temperature and near-true temperature measured by thermocouple sensor are provided, where (a) is the scatter plot of uncompensated temperature and near-true temperature measured by thermocouple sensor, and (b) is the scatter plot of compensated temperature and near-true temperature measured by thermocouple sensor.

[0040] Figure 7 Scatter plots of uncompensated temperature, compensated temperature and near-true temperature measured by fiber optic sensor are provided, where (a) is the scatter plot of uncompensated temperature and near-true temperature measured by fiber optic sensor and (b) is the scatter plot of compensated temperature and near-true temperature measured by fiber optic sensor. Detailed Implementation

[0041] To better understand the present invention, specific embodiments are described below. It should be noted that the following embodiments are for further illustration only and should not be construed as limiting the scope of protection of the present invention. Non-essential improvements and adjustments made by those skilled in the art based on the above description of the present invention still fall within the scope of protection of the present invention.

[0042] Example

[0043] Based on the temperature data tracked and measured under various temperature measuring elements and parameters during the microwave heating process, the present invention will be further described in detail below with reference to embodiments.

[0044] Step 1: Using thermocouples, infrared sensors, and fiber optic cables, the temperature changes of the material during microwave heating are tracked and measured in real time under different key parameter conditions. Data is acquired once per second. Taking a power density of 35 W / g as an example, the temperature data measured by thermocouples, infrared sensors, and fiber optic cables at different heating times are shown in Table 1. Due to the large amount of temperature data read, Table 1 presents data examples at 20-second intervals. Further calculations are then performed to determine the first-order difference between the temperature measurements of each sensor at adjacent time points under these parameter conditions. The results are shown in Table 2. Tables 1 and 2 are for illustrative purposes only. Due to the large amount of data, the temperature values ​​of other temperatures and parameters are omitted. The calculation results will be presented directly below.

[0045] Table 1. Temperature data measured by three sensors at a power density of 35 W / g.

[0046]

[0047]

[0048] Table 2. First-order difference of temperature measured by three sensors at a power density of 35 W / g

[0049]

[0050]

[0051] Step 2: Calculate the fluctuation and negative change ratio based on the temperature after first-order difference analysis of different sensors, and then calculate the stability score of different sensors based on the fluctuation and negative change ratio.

[0052] ,

[0053] ,

[0054] ,

[0055] In the formula, This indicates the magnitude of temperature fluctuation in the data measured by each temperature sensor, in μ. k Indicates the average range of change. This represents the proportion of negative changes, i.e., the frequency of temperature decreases during the heating process; N represents the total number of changes. This represents a score indicating the stability of temperature measurement data from different sensors. Specifically, based on a weighted calculation method, the volatility of different sensors under different measurement conditions and parameters is first calculated. ) and the proportion of negative changes ( The results are shown in Tables 3-5.

[0056] Table 3. Proportion of fluctuations and negative changes of infrared thermometers under different measurement conditions and parameters.

[0057] Table 4. Proportion of fluctuations and negative changes in thermocouple temperature sensing elements under different measurement conditions and parameters.

[0058] Table 5. Proportion of fluctuation and negative change of fiber optic temperature sensing element under different measurement conditions and parameters.

[0059]

[0060] Step 3: Based on the above calculation results, calculate the fluctuation of different sensors under all measurement conditions. ) and the proportion of negative changes ( The average value of ) is used as a representative, and then the stability score of each sensor is further calculated based on this average value. The results of the normalized weights and normalized weights are shown in Table 6.

[0061] ,

[0062] ,

[0063] In the formula, The score represents the stability calculation of temperature measurement data from different sensors. This represents the normalized weight of temperature measurements from each temperature sensor.

[0064] Table 6. Stability scores and weights of different sensors

[0065]

[0066] Step 4: Calculate the weighted fusion temperature based on the normalized weights of different sensors. The formula is as follows:

[0067] .

[0068] The sensor temperatures and weighted fusion temperatures under different measurement conditions and parameters were calculated using this formula. Changes such as Figures 1-4 As shown in the example. The horizontal axis of the graph represents the heating time, and the vertical axis represents the temperature values ​​measured by different sensors during that heating time, as well as the calculated weighted fusion temperature. Figures 1-4 It can be seen that, compared with the temperature change curve of a single sensor, the temperature change curve of the weighted fusion temperature is more stable and can better reflect the near-real temperature change of the material during the heating process.

[0069] Step 5: To achieve near-true temperature prediction of materials using a single sensor, linear regression, quadratic polynomial regression, and cubic polynomial regression equations are used for fitting. The data is randomly divided into training and validation sets in a 7:3 ratio. The root mean square error (RMSEC) of the training set and the coefficient of determination (CDO) are used as the validation set parameters. ), Root Mean Square Error of Validation Set (RMSEP), Coefficient of Determination of Validation Set ( The model was evaluated, and the results are shown in Table 7. The models with lower RMSEP or lower RMSEP were then selected. The one with the higher value is the optimal algorithm.

[0070] Table 7 Comparison of Evaluation Results of Different Algorithms

[0071]

[0072] Step 6: Extract algorithm parameters and integrate the models.

[0073] T = 0.8721429 × infrared^3 + 0.0006499 × infrared^2 - 0.0000009 × infrared + 5.9375739,

[0074] T = 1.1295455 × thermocouple^3 - 0.0006251 × thermocouple^2 + 0.0000007 × thermocouple - 4.2043621,

[0075] T = 0.9923852 × fiber^2 + 0.0000069 × fiber + 1.2544064.

[0076] The above temperature compensation model is used to compensate for the temperatures measured by infrared, thermocouple, and fiber optic sensors at different power densities. The uncompensated temperature, compensated temperature, and approximate true temperature are measured by each sensor. The scatter plot results for temperature are as follows: Figures 5-7 As shown, the average absolute error of the temperature measured by the infrared sensor decreased from 5.701℃ before compensation to 3.027℃ after compensation; the average absolute error of the temperature measured by the thermocouple sensor decreased from 4.749℃ before compensation to 3.38℃ after compensation; and the average absolute error of the temperature measured by the fiber optic sensor decreased from 2.303℃ before compensation to 1.421℃ after compensation. Their measurement accuracies improved by 46.904%, 28.827%, and 38.298%, respectively, further verifying that compensation effectively improves the measurement accuracy of various temperature sensors.

[0077] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or equivalent modifications to the above-disclosed technical content. However, any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for dynamic measurement and compensation of microwave heating temperature based on multi-source data fusion, characterized in that, Includes the following steps: Step 1: Using different types of temperature sensors, measure the dynamic temperature data of the material during the process of multiple parameter changes under microwave heating, and then calculate the first-order difference between adjacent time points based on the temperature data from each sensor. This refers to the magnitude of temperature change during the heating process; a positive number indicates that the temperature has increased compared to the previous measurement, while a negative number indicates that the temperature has decreased compared to the previous measurement. In the formula ， This represents the temperature measurement value of the k-th sensor at time point i. This indicates the magnitude of temperature change during the heating process. k Indicates the sensor type, i Indicates a time index; Step 2: Calculate the fluctuation based on the temperature after first-order difference analysis from different sensors. and the proportion of negative changes : ， ， In the formula ， This indicates the magnitude of temperature fluctuations in the temperature data measured by each temperature sensor. μ k Indicates the data passed through the k-th sensor. The calculated mean temperature change This represents the proportion of negative changes, that is, the frequency of temperature decreases during the heating process. N Indicates the number of times the total change occurs; Step 3: Calculate the stability score of temperature measurement data from different sensors based on the proportion of volatility and negative change. Then, the normalized weights of each sensor are calculated to ensure that the sum of the weights of all sensors is 1. ， In the formula ， This represents the score calculated based on the stability of temperature measurement data from different sensors. This indicates the normalized weighting of temperature measurements from each temperature sensor. Step 4: Calculate the weighted fusion temperature as the final temperature based on the normalized weights of each temperature sensor and its corresponding temperature value. In the formula ， This represents the approximate true temperature data calculated based on temperature measurement data from various temperature sensors and normalized weights. Step 5: Using the temperature obtained from each temperature sensor as the independent variable and the weighted fusion temperature as the dependent variable, a regression analysis is performed using a regression algorithm to construct a temperature measurement compensation model for each temperature sensor. The temperature measurement compensation model is used to accurately correct the original temperature measurement value of the sensor.

2. The method according to claim 1, characterized in that, The temperature sensors include thermocouple sensors, infrared sensors, and fiber optic sensors.

3. The method according to claim 1, characterized in that, In step one, the parameters include sample heating temperature, sample placement position, initial sample moisture content, and input power density.

4. The method according to claim 1, characterized in that, In step five, the regression analysis includes linear regression, quadratic polynomial regression, or trigonometric polynomial regression.

5. The method according to claim 4, characterized in that, Temperature measurement data under different parameter conditions were randomly divided into training and validation sets in a 7:3 ratio. The root mean square error (RMSEC) of the training set and the coefficient of determination of the training set were used as the validation set parameters. Root mean square error of validation set (RMSEP) and coefficient of determination of validation set Evaluate the model.

6. The method according to claim 1, characterized in that, The process also includes step six: based on the actual sensor conditions of the microwave heating equipment, embed the compensation model corresponding to each temperature sensor into the PLC program control system, and correct the original measurement values ​​of the sensors by calling the compensation model in real time, and output the corrected temperature data to the equipment control module to achieve closed-loop precise control of microwave heating temperature.

7. The method according to claim 1, characterized in that, The interval between adjacent time points is a time length selected between 1 and 5 seconds.

8. The method according to claim 7, characterized in that, The interval between adjacent time points is 1 second.

Images