TWI870160B - System and method for estimating product simulation parameter - Google Patents

Abstract

A system and a method for estimating a product simulation parameter are provided. The method includes following steps: obtaining a product thermal image corresponding to a product, and obtaining a product temperature when the product is in a tested oven, wherein an equipment information, the product thermal image and the product temperature correspond to a same time point; and utilizing the product thermal image, the product temperature, a product information and the equipment information to obtain a product virtual coefficient corresponding to the product, and using the product virtual coefficient to estimate a product simulation parameter when the product is in a to-be-tested-oven.

Description

System and method for estimating product simulation parameters

The present invention relates to a system and method for estimating product simulation parameters.

In the field of thermal processing technology, usually only a small number of sensors are installed in some areas of the thermal processing equipment, resulting in the inability to obtain product parameters (such as temperature and moisture content) when the product is in different ovens in the thermal processing equipment. If sensors are widely installed, more information can be obtained, but the cost will be greatly increased due to the high price of sensors. Therefore, in the field of thermal processing technology, usually only the experience of personnel can be relied on to adjust the thermal processing equipment to ensure product quality.

The system for estimating product simulation parameters of the present invention is used for thermal processing equipment and products, wherein the thermal processing equipment includes a tested oven and at least one oven to be tested, wherein the product corresponds to product information, wherein the thermal processing equipment corresponds to equipment information, wherein the system includes a first thermal imaging sensor, a second thermal imaging sensor, a temperature sensor, a storage medium and a processor. The first thermal imaging sensor is arranged at the inlet section of the thermal processing equipment. The second thermal imaging sensor is arranged at the outlet section of the thermal processing equipment. The temperature sensor is arranged at the tested oven. The storage medium stores a sensing data acquisition unit and a first product simulation parameter estimation unit, wherein the sensing data acquisition unit is used to obtain a product thermal image corresponding to the product through a first thermal image sensor and a second thermal image sensor, and to obtain the product temperature of the product in the tested oven through a temperature sensor, wherein the equipment information, the product thermal image and the product temperature correspond to the same time point; the first product simulation parameter estimation unit is used to obtain a product virtual coefficient corresponding to the product using the product thermal image, the product temperature, the product information and the equipment information, and to estimate a first product simulation parameter of the product in the tested oven using the product virtual coefficient. The processor is coupled to the first thermal image sensor, the second thermal image sensor, the temperature sensor and the storage medium, and is used to access and execute the sensing data acquisition unit and the first product simulation parameter estimation unit.

The method for estimating product simulation parameters of the present invention is used for thermal processing equipment and products, wherein the thermal processing equipment includes a tested oven and at least one oven to be tested, wherein the product corresponds to product information, wherein the thermal processing equipment corresponds to equipment information, and wherein the method includes the following steps: obtaining a product thermal image corresponding to the product, and obtaining the product temperature of the product in the tested oven, wherein the equipment information, the product thermal image and the product temperature correspond to the same time point; and using the product thermal image, the product temperature, the product information and the equipment information to obtain a product virtual coefficient corresponding to the product, and using the product virtual coefficient to estimate a first product simulation parameter of the product in the oven to be tested.

FIG. 1 is a schematic diagram of a system 100 for estimating product simulation parameters according to an embodiment of the present invention. Please refer to FIG. 1 . It should be noted that the system 100 for estimating product simulation parameters can be used for thermal processing equipment and products, and can further estimate the simulation parameters of the product (such as temperature and moisture content). In detail, the thermal processing equipment may include a tested oven and at least one oven to be tested, and the oven to be tested may be the same oven as the tested oven, and there is no limitation here. In addition, the product may correspond to product information. On the other hand, the thermal processing equipment may correspond to equipment information. In this embodiment, the system 100 for estimating product simulation parameters may include a first thermal image sensor 110, a second thermal image sensor 120, a temperature sensor 130, a storage medium 140, a processor 150, and a moisture content sensor 160. The first thermal image sensor 110 may be disposed at an inlet section of a thermal processing device. The second thermal image sensor 120 may be disposed at an outlet section of the thermal processing device. The temperature sensor 130 may be disposed in a measured oven. The storage medium 140 may store a sensing data acquisition unit 141, a first product simulation parameter estimation unit 142, and a second product simulation parameter estimation unit 143. The processor 150 may be coupled to the first thermal image sensor 110, the second thermal image sensor 120, the temperature sensor 130, the storage medium 140, and the moisture sensor 160, and access and execute the sensing data acquisition unit 141, the first product simulation parameter estimation unit 142, and the second product simulation parameter estimation unit 143. The moisture sensor 160 may be disposed at the outlet section (of the thermal processing equipment). In other embodiments, the system 100 for estimating product simulation parameters may omit the moisture sensor 160 and the second product simulation parameter estimation unit 143.

FIG2 is a flow chart of a method for estimating product simulation parameters according to an embodiment of the present invention, wherein the method can be implemented by the system 100 for estimating product simulation parameters shown in FIG1. Please refer to FIG1 and FIG2 at the same time.

First, execute step S210 to obtain a product thermal image corresponding to the product, and obtain the product temperature of the product in the measured oven, wherein the equipment information, the product thermal image, and the product temperature correspond to the same time point. Specifically, in step S210, the sensing data acquisition unit 141 can obtain a product thermal image corresponding to the product through the first thermal image sensor 110 and the second thermal image sensor 120, and can obtain the product temperature of the product in the measured oven through the temperature sensor 130, wherein the equipment information, the product thermal image, and the product temperature correspond to the same time point. In other words, the equipment information, product thermal image, and product temperature obtained at the same time point can be used in the subsequent steps of the present invention after being integrated/mapped.

Then, step S230 is executed to obtain a product virtual coefficient corresponding to the product using the product thermal image, product temperature, product information, and equipment information, and to estimate the first product simulation parameter of the product in the oven to be tested using the product virtual coefficient. Specifically, in step S230, the first product simulation parameter estimation unit 142 can obtain a product virtual coefficient corresponding to the product using the product thermal image, product temperature, product information, and equipment information, and to estimate the first product simulation parameter of the product in the oven to be tested using the product virtual coefficient. In this embodiment, the first product simulation parameter is the simulated temperature of the product. The following will describe an implementation example of "obtaining a product virtual coefficient corresponding to the product" in step S230.

FIG3A is an implementation example of "obtaining a product virtual coefficient corresponding to a product" in step S230 shown in FIG2. Please refer to FIG1, FIG2 and FIG3A at the same time. In this embodiment, the first product simulation parameter estimation unit 142 can use the product thermal image, product temperature, product information and equipment information to convert the energy conservation equation Convert to ordinary differential equation In one embodiment, the equipment information may include the measured oven temperature corresponding to the measured oven, and the equipment information may include the measured oven temperature corresponding to the oven to be tested. In other embodiments, the equipment information may also include but is not limited to process parameters, host speed, and exhaust speed. In one embodiment, the product information may include but is not limited to fabric, code weight, additives, width, density, specific heat capacity, and thickness.

Then, the first product simulation parameter estimation unit 142 can analyze the ordinary differential equation to obtain the iterative equation ( - ). Then, the first product simulation parameter estimation unit 142 may use the product information and the equipment information to perform Bayesian optimization fitting on the product virtual coefficient. Then, the first product simulation parameter estimation unit 142 may substitute the product information, the equipment information and the product virtual coefficient into the iterative equation to obtain the first product simulation parameter. In one embodiment, the above-mentioned Bayesian optimization fitting may include the following steps (a), (b) and (c).

(a) The first product simulation parameter estimation unit 142 uses an iterative equation to solve the first product simulation parameter, wherein the first product simulation parameter includes a first product simulation parameter minimum value.

(b) The first product simulation parameter estimating unit 142 updates the product virtual coefficient using the difference between the product temperature and the minimum value of the first product simulation parameter.

(c) The first product simulation parameter estimation unit 142 determines whether the difference between the product temperature and the minimum value of the first product simulation parameter corresponds to the minimum average absolute error percentage value, wherein if the judgment result is "no", step (b) is repeated; otherwise, if the judgment result is "yes", the product virtual coefficient is obtained. .

FIG3B is another implementation example of “obtaining a product virtual coefficient corresponding to the product” in step S230 shown in FIG2 . Please refer to FIG1 , FIG2 and FIG3B at the same time. In this embodiment, the product and the heat treatment equipment may correspond to a heating stage. Specifically, the heating stage may include a first stage (product temperature is less than T _transit ), a second stage (product temperature reaches T _transit ) and a third stage (product temperature exceeds T _transit ). T _transit is, for example, 100° C. In other words, the above-mentioned heating stages are different stages of heating the product in the heat treatment equipment. The first product simulation parameter estimation unit 142 can obtain the product virtual coefficient by utilizing the energy supply, mass change, and temperature difference change of the product during the heating stage.

In this embodiment, it is assumed that T _transit is the evaporation temperature, T _product is the product temperature, X ₀ is the initial moisture content of the product, m _total is the inlet total mass of the product, and u is the speed of the product in the thermal processing equipment. As shown in Formula 1, the moisture content (MC) can be obtained based on the moisture regain (MR), for example, the moisture content can be 40%. As shown in Formula 2, the mass of the product (m _product ) can be obtained through the outlet weight of the product (m _final ), the initial moisture content of the product (X ₀ ) and the outlet moisture content of the product (X ₁ ). As shown in Formula 3, the inlet total mass of the product (m _total ) can be obtained from the initial moisture content of the product (X ₀ ) and the mass of the product (m _product ). Before the moisture content of the product reaches the boiling point (T _transit ), the product and moisture coexist, so the first product simulation parameter estimation unit 142 can use the method of mixed specific heat to estimate the theoretical value of heat transfer of the actual product in the oven. The calculation method of the mixed density (ρ _mix ) can be shown in Formula 4. On the other hand, the calculation method of the mixed specific heat (Cp _mix ) can be shown in Formula 5. The calculation method of the Reynolds number (Re) can be shown in Formula 6, where ρ _a is the hot air density, V _jet is the pore jet flow rate, D is the pore diameter, and μ is the dynamic viscosity. The calculation method of the Nusselt number (Nu) can be shown in Formula 7, where Pr is the Prandtl number, H is the distance from the pore to the product, D is the pore diameter, f is the relative pore area, and Re is the Reynolds number. The heat transfer coefficient (h) can be calculated as shown in Formula 8, where Nu is the Newton number, k is the thermal conductivity, and D is the pore diameter. … (Formula 1) … (Formula 2) … (Formula 3) … (Formula 4) … (Formula 5) where ρ _mix is the mixture density, m _wi is the inlet water weight, ρ _w is the density of water, ρ _product is the density of the product fabric, C _{p_mix} is the mixture specific heat, C _{p_w} is _the specific heat of water, and C _{p_product} is the specific heat of the product. … (Formula 6) … (Formula 7) … (Formula 8)

Furthermore, the convection heat (Q) can be calculated as in Formula 9, where h is the heat transfer coefficient, A is the fabric area, T _oven is the measured oven temperature, and T _product is the product temperature. The radiation heat (Q _rad ) can be calculated as in Formula 10, where ε is the emissivity, σ is the Stefan-Boltzmann constant, T _oven is the measured oven temperature, and T _product is the product temperature. Temperature change ( ) can be calculated as shown in Formula 11. The amount of water removed (m _del ) can be calculated as shown in Formula 12, where h _fg is the latent heat of evaporation of water. … (Formula 9) … (Formula 10) … (Formula 11) … (Formula 12)

After the first product simulation parameter estimation unit 142 (using the implementation of FIG. 3A or FIG. 3B ) obtains the product virtual coefficient corresponding to the product, the first product simulation parameter estimation unit 142 can use the product virtual coefficient to estimate the first product simulation parameter of the product in the oven to be tested.

In one embodiment, the product may include a printed circuit board (PCB).

In one embodiment, the product may include fabric. After the first product simulation parameter estimation unit 142 completes step S230 of FIG. 2 (also known as the step of estimating the first product simulation parameter using the "physical model"), the second product simulation parameter estimation unit 143 may then estimate the second product simulation parameter corresponding to the product. This will be described further below.

FIG. 4A is a schematic diagram of estimating the second product simulation parameter according to an embodiment of the present invention. FIG. 4B is a further explanation of FIG. 4A. Please refer to FIG. 1, FIG. 2, FIG. 4A and FIG. 4B simultaneously. In this embodiment, the second product simulation parameter estimation unit 143 can input the first product simulation parameter, product information and equipment information into the neural model to establish a two-dimensional second product simulation parameter distribution model (for example, a two-dimensional moisture content distribution model, which means that in this embodiment, the second product simulation parameter corresponding to the product is estimated to be the moisture content). In one embodiment, the neural model may include 5 network layers, and can be normally assigned as a weight initialization (Kernel initializer). Furthermore, the neural model can use a linear rectifier function (Rectified Linear Unit, ReLU) as an activation function, and the batch size of each update is 150, the epochs is 128, and the root mean square error can be used as a loss function. In this embodiment, the sensing data acquisition unit 141 can obtain the product moisture content corresponding to the product through the moisture sensor 160. Then, the second product simulation parameter estimation unit 143 can input the first product simulation parameter and the product moisture content into the two-dimensional second product simulation parameter distribution model to estimate the second product simulation parameter corresponding to the product.

For example, as shown in FIG4B , it is assumed that the first product simulation parameter includes _Ti1 , _Ti2 , and _Ti3 , and the first product simulation parameter _Ti1 , the first product simulation parameter _Ti2 , and the first product simulation parameter _Ti3 correspond to pixel 1, pixel 2, and pixel 3, respectively. The second product simulation parameter estimation unit 143 may input the first product simulation parameter _Ti1 and the product moisture content (obtained by the moisture content sensor 160 disposed at the outlet section of the thermal processing equipment) into the two-dimensional second product simulation parameter distribution model to estimate the second product simulation parameter _Mo1 of pixel 1. Similarly, the second product simulation parameter estimation unit 143 can input the first product simulation parameter _Ti2 and the product moisture content (obtained by the moisture content sensor 160 disposed at the outlet section of the heat treatment equipment) into the two-dimensional second product simulation parameter distribution model to estimate the second product simulation parameter _Mo2 of the pixel 2. Similarly, the second product simulation parameter estimation unit 143 can input the first product simulation parameter _Ti3 and the product moisture content (obtained by the moisture content sensor 160 disposed at the outlet section of the heat treatment equipment) into the two-dimensional second product simulation parameter distribution model to estimate the second product simulation parameter _Mo3 of the pixel 3. Next, the second product simulation parameter estimating unit 143 may utilize the second product simulation parameter _Mo1 , the second product simulation parameter _Mo2 , and the second product simulation parameter _Mo3 to obtain a second product simulation parameter corresponding to the product.

In summary, after obtaining the product virtual coefficient corresponding to the product, the system and method of the present invention for estimating product simulation parameters can use the product virtual coefficient to estimate the first product simulation parameter of the product in the oven to be tested. In particular, the present invention can use product thermal images, product temperature, product information and equipment information to obtain product virtual coefficients, so that the first product simulation parameter can be estimated more accurately (based on the "physical model"). Furthermore, the system and method of the present invention for estimating product simulation parameters can also use the first product simulation parameter to estimate the second product simulation parameter. Based on this, the convenience of adjusting thermal processing equipment can be significantly improved.

100: System for estimating product simulation parameters

110: The first thermal imaging sensor

120: Second thermal imaging sensor

130: Temperature sensor

140: Storage Media

141: Sensor data acquisition unit

142: First product simulation parameter estimation unit

143: Second product simulation parameter estimation unit

150:Processor

160: Moisture content sensor

S210, S230: Step

T _i1 , T _i2 , T _i3 : First product simulation parameters

_Mo1 , _Mo2 , _Mo3 : Second product simulation parameters

1, 2, 3: Pixels

FIG. 1 is a schematic diagram of a system for estimating product simulation parameters according to an embodiment of the present invention. FIG. 2 is a flow chart of a method for estimating product simulation parameters according to an embodiment of the present invention. FIG. 3A is an implementation example of "obtaining a product virtual coefficient corresponding to a product" in step S230 shown in FIG. 2. FIG. 3B is another implementation example of "obtaining a product virtual coefficient corresponding to a product" in step S230 shown in FIG. 2. FIG. 4A is a schematic diagram of estimating a second product simulation parameter according to an embodiment of the present invention. FIG. 4B is a further explanation of FIG. 4A.

S210, S230: Steps

Claims (15)

A system for estimating product simulation parameters, used for thermal processing equipment and products, wherein the thermal processing equipment includes a tested oven and at least one oven to be tested, wherein the product corresponds to product information, wherein the thermal processing equipment corresponds to equipment information, wherein the system includes: a first thermal imaging sensor, arranged at an inlet section of the thermal processing equipment; a second thermal imaging sensor, arranged at an outlet section of the thermal processing equipment; a temperature sensor, arranged at the tested oven; a storage medium, storing a sensing data acquisition unit and a first product simulation parameter estimation unit, wherein the The sensing data acquisition unit is used to obtain a product thermal image corresponding to the product through the first thermal image sensor and the second thermal image sensor, and to obtain the product temperature of the product in the measured oven through the temperature sensor, wherein the equipment information, the product thermal image and the product temperature correspond to the same time point; the first product simulation parameter estimation unit is used to obtain a product virtual coefficient corresponding to the product using the product thermal image, the product temperature, the product information and the equipment information, and to obtain a product virtual coefficient corresponding to the product using the product thermal image, the product temperature, the product information and the equipment information. The system further comprises a moisture content sensor coupled to the processor, wherein the moisture content sensor is disposed at the outlet section, wherein the product comprises fabric, and wherein the storage medium further stores a second product simulation parameter estimation unit. , wherein the second product simulation parameter estimation unit is used to input the first product simulation parameter, the product information and the equipment information into the neural model to establish a two-dimensional second product simulation parameter distribution model; the sensing data acquisition unit is used to obtain the product moisture content corresponding to the product through the moisture content sensor; the second product simulation parameter estimation unit is used to input the first product simulation parameter and the product moisture content into the two-dimensional second product simulation parameter distribution model to estimate the second product simulation parameter corresponding to the product. The system as described in claim 1, wherein the first product simulation parameter estimation unit is used to convert the energy conservation equation into an ordinary differential equation using the product thermal image, the product temperature, the product information, and the equipment information; the first product simulation parameter estimation unit is used to analyze the ordinary differential equation to obtain an iterative equation; the first product simulation parameter estimation unit is used to perform Bayesian optimization fitting on the product virtual coefficient using the product information and the equipment information; the first product simulation parameter estimation unit is used to substitute the product information, the equipment information, and the product virtual coefficient into the iterative equation to obtain the first product simulation parameter. The system as described in claim 2, wherein the Bayesian optimization fitting includes the following steps: the first product simulation parameter estimation unit uses the iterative equation to solve the first product simulation parameter, wherein the first product simulation parameter includes the first product simulation parameter minimum value; and the first product simulation parameter estimation unit uses the difference between the product temperature and the first product simulation parameter minimum value to update the product virtual coefficient; and the first product simulation parameter estimation unit determines whether the difference between the product temperature and the first product simulation parameter minimum value corresponds to the minimum mean absolute error percentage value to obtain the product virtual coefficient. A system as described in claim 1, wherein the equipment information includes a measured oven temperature corresponding to the measured oven, and the equipment information includes a measured oven temperature corresponding to the oven to be measured. A system as described in claim 1, wherein the product and the heat treatment equipment correspond to a heating stage, wherein the first product simulation parameter estimation unit is used to obtain the product virtual coefficient by utilizing the energy supply, mass change and temperature difference change of the product during the heating stage. A system as described in claim 1, wherein the product information includes fabric, code weight, additives, width, density, specific heat capacity, and thickness. A system as described in claim 1, wherein the product includes a printed circuit board (PCB). A method for estimating product simulation parameters, for thermal processing equipment and products, wherein the thermal processing equipment includes a tested oven and at least one oven to be tested, wherein the product corresponds to product information, wherein the thermal processing equipment corresponds to equipment information, wherein the method comprises the following steps: obtaining a product thermal image corresponding to the product, and obtaining a product temperature of the product when the product is in the tested oven, wherein the equipment information, the product thermal image and the product temperature correspond to the same time point; and using the product thermal image, the product temperature, the product information and the equipment information to A product virtual coefficient corresponding to the product is obtained, and the product virtual coefficient is used to estimate the first product simulation parameter of the product in the oven to be tested, wherein the method further includes the following steps: inputting the first product simulation parameter, the product information and the equipment information into a neural model to establish a two-dimensional second product simulation parameter distribution model; obtaining the product moisture content corresponding to the product; and inputting the first product simulation parameter and the product moisture content into the two-dimensional second product simulation parameter distribution model to estimate the second product simulation parameter corresponding to the product. The method of claim 8, wherein the step of using the product thermal image, the product temperature, the product information and the equipment information to obtain the product virtual coefficient corresponding to the product includes: using the product thermal image, the product temperature, the product information and the equipment information to convert the energy conservation equation into an ordinary differential equation; analyzing the ordinary differential equation to obtain an iterative equation; using the product information and the equipment information to perform Bayesian optimization fitting on the product virtual coefficient; substituting the product information, the equipment information and the product virtual coefficient into the iterative equation to obtain the first product simulation parameter. As described in claim 9, the Bayesian optimization fitting includes the following steps: using the iterative equation to solve the first product simulation parameter, wherein the first product simulation parameter includes the first product simulation parameter minimum value; using the difference between the product temperature and the first product simulation parameter minimum value to update the product virtual coefficient; and determining whether the difference between the product temperature and the first product simulation parameter minimum value corresponds to the minimum mean absolute error percentage value to obtain the product virtual coefficient. The method as claimed in claim 8, wherein the equipment information includes the measured oven temperature corresponding to the measured oven, and the equipment information includes the measured oven temperature corresponding to the oven to be measured. The method of claim 8, wherein the product and the heat treatment equipment correspond to a heating stage, wherein the step of using the product thermal image, the product temperature, the product information and the equipment information to obtain the product virtual coefficient corresponding to the product includes: using the energy supply, mass change and temperature difference change of the product during the heating stage to obtain the product virtual coefficient. A method as claimed in claim 8, wherein the product comprises a fabric. The method as claimed in claim 8, wherein the product information includes fabric, code weight, additives, width, density, specific heat capacity and thickness. A method as described in claim 8, wherein the product includes a printed circuit board.