JP2020201071A - Method for predicting temperature, temperature prediction device, and temperature prediction program - Google Patents

Abstract

To provide a method for predicting a temperature which can reduce the separation between a predicted temperature and the actual temperature of a measurement target and can finish a prediction in a shorter time.SOLUTION: A predicted value (Te1,Te2, omitted, Ten) predicted by each prediction model is calculated by using n types (n≥1) of prediction models on the basis of an actually measured temperature of a measurement target obtained by a temperature sensor (Tactually) and a change per unit time in the actually measured temperature (Tactually) (S5). By using the relational expression between the actually measured temperature (Tactually) obtained in advance, the predicted temperatures (Te1,Te2, omitted, Ten), and a predicted temperature (Tpredicted) of the measurement target, the temperature (Tpair) of the measurement target is predicted as the predicted temperature (Tpredicted) before the actually measured temperature (Tactually) of the temperature sensor becomes equal to the temperature (Tpair) of the measurement target (S6).SELECTED DRAWING: Figure 13

Description

The present invention relates to a temperature prediction method, a temperature prediction device, and a temperature prediction program for predicting the temperature of a measurement target based on the measured temperature of a temperature sensor in contact with the measurement target.

For example, bearings that are often used in rotating machines generate heat when an abnormality such as poor lubrication occurs and raise the temperature of the bearing box or the like, so temperature measurement is performed to monitor the state of the bearing. When the temperature sensor is brought into contact with the measurement target in order to measure the temperature of the measurement target such as a bearing, the temperature sensor changes (rises or falls) to the temperature of the measurement target because there is a temperature difference between the measurement target and the temperature sensor. ). The time required for the temperature sensor to change to the temperature to be measured is the time required for temperature measurement, but if the thermal capacity of the temperature sensor cannot be sufficiently reduced due to the structural restrictions of the temperature sensor, the time required for temperature measurement is It cannot be ignored. In particular, industrial temperature sensors used in various environments are required to have structural robustness, which may slow down the temperature response. Therefore, the temperature after the temperature sensor is brought into contact with the measurement target. By predicting the predicted temperature of the measurement target based on the change in the measured temperature of the sensor, the time required for temperature measurement is shortened.

Newton's law of cooling has been conventionally known as this type of temperature prediction method. Using Newton's law of cooling, the temperature to be measured can be predicted from the prediction model of the following equation (6).

Here, t is time, T (t) is the measured temperature at time t, a is a coefficient, and _Te is the predicted temperature of the measurement target. The measured temperature T (t) is the measured temperature of the temperature sensor in contact with the measurement target.

Newton's law of cooling uses an empirical rule that the change dT (t) / dt of the measured temperature per unit time is proportional to the difference between the measured temperature T (t) and the temperature to be measured.

As a prediction method other than Newton's law of cooling, for example, the prediction method described in Patent Document 1 is known. This prediction method is used for thermometers. In this prediction method, the starting point of the prediction calculation is when the measured value by the temperature sensor is equal to or higher than the predetermined value and the temperature rise rate is equal to or higher than the predetermined value, the predicted value is Y, the measured value is T, and the multiplication amount is U. Then, the prediction model given by Y = T + U is used. The addition amount is given by U = a ₁ × dT / dt + b ₁ or U = (a ₂ × t + b ₂ ) × dT + (c ₂ × t + d ₂ ), where t is the elapsed time from the starting point. Here, the coefficients a ₁ , b ₁ , a ₂ , b ₂ , c ₂ , and d ₂ are determined based on the characteristics of the subject and the characteristics of the temperature sensor.

Japanese Unexamined Patent Publication No. 2007-24530

However, the former prediction method based on Newton's law of cooling has a problem that the difference between the predicted temperature and the temperature to be measured is large, as shown in FIG. In particular, there is a large discrepancy between the predicted temperature immediately after the temperature sensor comes into contact with the measurement target and the temperature of the measurement target.

Further, in the prediction method used for the latter thermometer, it is possible to make a highly accurate prediction, but there is a problem that it takes time to complete the prediction.

Therefore, the present invention provides a temperature prediction method, a temperature prediction device, and a temperature prediction program that can reduce the difference between the predicted temperature of the measurement target and the temperature of the measurement target and can complete the prediction in a short time. The purpose is to provide.

In order to solve the above problem, one aspect of the present invention is a temperature prediction method for predicting the temperature (T _pair ) of the measurement target based on the measured temperature (T _actual ) of the temperature sensor in contact with the measurement target. Based on the temperature change of the measured temperature (T _real ) and the measured temperature (T _real ) of the measurement target by the temperature sensor per unit time, n kinds (n ≧ 1) of prediction models are used, and each prediction model is used. predicted value _{(T e1, T e2, ...} T en) calculates, previously obtained actual temperature and (T _actual) predicted value _{(T e1, T e2, ...} T en) and measured the predicted temperature (T using a relational expression between _pre) predicts the temperature of the measurement target (T _pairs) before measured temperature of the temperature sensor (T _actual) is equal to the temperature (T _pairs) to be measured as a predicted temperature (T _pre) It is a temperature prediction method.

Another aspect of the present invention is a temperature prediction device that predicts the temperature (T _pair ) of the measurement target based on the measured temperature (T _actual ) of the temperature sensor in contact with the measurement target, and the measurement target by the temperature sensor. Based on the temperature change per unit time of the measured temperature (T _real ) and the measured temperature (T _real ), n kinds (n ≧ 1) of prediction models are used, and the predicted values (The ₁ , T) by each prediction model are used. _e2, ... and the prediction value calculation means for calculating the _{T en),} previously obtained actual temperature and (T _actual) predicted value _{(T e1, T e2, ...} T en) and measured the predicted temperature (T _pre) using a relational expression between the temperature to predict the temperature of the measurement target (T _pairs) before measured temperature of the temperature sensor (T _actual) is equal to the temperature (T _pairs) to be measured as a predicted temperature (T _pre) It is a temperature prediction device including a prediction means.

Yet another aspect of the present invention is a temperature prediction program that predicts the temperature (T _pair ) of the measurement target based on the measured temperature (T _actual ) of the temperature sensor in contact with the measurement target, and is a device having a processor. In addition, based on the temperature change of the measured temperature (T _real ) and the measured temperature (T _real ) of the measurement target by the temperature sensor per unit time, n kinds (n ≧ 1) of prediction models are used, and each prediction model is used. predicted value by _{(T e1, T e2, ...} T en) measured and predicted value calculation step of calculating, previously obtained actual temperature and (T _actual) predicted value _{(T e1, T e2, ...} T en) and using a relational expression between the target predicted temperature (T _pre), predicting the temperature sensor measured temperature (T _actual) temperature (T _pairs) to be measured prior to match the temperature (T _pairs) of the measurement target temperature a temperature prediction step of predicting the (T _pre), a temperature predicting program for executing.

According to the present invention, it is possible to reduce the deviation between the predicted temperature of the measurement target (T _pre) and temperature (T _pairs) to be measured, in a short time temperature prediction without being affected by the heat capacity of the temperature sensor Can be finished.

Measured temperature T _actual by the temperature sensor, the temperature T _versus measured, predicted value T _e1 with predictive models _A, is a graph comparing the predicted value T _e2 by the prediction model B. Measured temperature T _actual is a three-dimensional scatter diagram plotting the difference between the predicted value T _e2 by the prediction value T _e1, the temperature T _versus the measured prediction model B by the prediction model A. It is a two-dimensional scatter plot with γ = 1. Measured temperature T _actual, predicted temperature T _pre of the measuring object, the temperature T _pairs to be measured, a graph comparing the predicted value T _e1 by the prediction model A. Measured temperature T _actual, predicted temperature T _pre of the measuring object, the temperature T _pairs to be measured, a graph comparing the predicted value T _e1 by the prediction model A. Alpha by multiple regression analysis, beta, graph comparing the predicted temperature T _pre of the measuring object when called for gamma, alpha by a two-dimensional scatter diagram, beta, and predicted temperature T _pre of the measuring object when called for gamma Is. The predicted temperature T _pre using a predictive model A and the prediction model B, and a graph comparing the predicted temperature T _pre using a predictive model D and the prediction model C. The predicted temperature T _pre using predictive models A, B, a prediction model A, B, graph comparing the predicted temperature T _pre using C. Predicted temperature T _pre using one type of prediction model A, is a graph comparing the predicted temperature T _pre of using the predicted temperature T _pre using one prediction model D, 2 kinds of prediction model A, the B. FIG. 9 is an enlarged graph of FIG. Prediction model A, the predicted temperature T _pre with B, prediction model C, the predicted temperature T _pre with D, prediction model A, the predicted temperature T _pre with B (no actual temperature), predictive model C, and D it is a graph comparing the predicted temperature T _pre (no actual temperature) used. It is a block diagram which shows the internal structure of the temperature prediction apparatus. It is a flowchart of the process executed by a temperature prediction apparatus. It is a graph comparing the predicted temperature by the conventional Newton's law of cooling and the temperature of the measurement target.

Hereinafter, the temperature prediction method, the temperature prediction device, and the temperature prediction program according to the embodiment of the present invention will be described in detail based on the accompanying drawings. However, the present invention can be embodied in various forms and is not limited to the embodiments described herein. The present embodiment is provided with the intention of allowing those skilled in the art to fully understand the scope of the invention by adequately disclosing the specification.
<Temperature prediction method>
(Temperature prediction method using prediction model A and prediction model B)

The temperature prediction process of the present embodiment, first, by using the n types of predictive models, the predicted value by the respective prediction model T _{e1, T} e2, calculates a _... T _en. Here, using the prediction model A using Newton's law of cooling shown in equation (3) and the prediction model B using the empirical rule shown in equation (4) found by the inventor, the respective prediction models A and calculates the predicted value _{T e1, T e2} by B.

ここで、ｔは時間、Ｔ（ｔ）は時間ｔでの実測温度、ａは係数、Ｔ_ｅ１は予測モデルＡによる予測値である。

Here, t is time, T (t) is the measured temperature at time t, a is a coefficient, and _Te1 is a predicted value by the prediction model A.

By measuring the set of the measured temperature T (t) and the temperature change dT (t) / dt per unit time a plurality of times, the predicted value _Te1 and the coefficient a in the equation (3) can be estimated.

FIG. 1 shows a graph comparing the actual temperature measured by the temperature sensor, the temperature to be measured, and the predicted value _Te1 by the prediction model A. As shown in FIG. 1, the predicted value _Te1 by the prediction model A has a large deviation from the temperature of the measurement target. In particular, the dissociation immediately after the temperature sensor is brought into contact with the measurement target is remarkable.

The prediction model B shown in the equation (4) is a model of the actually obtained temperature response curve, and is based on an empirical rule.

ここで、ｔは時間、Ｔ（ｔ）は時間ｔでの実測温度、ｂは係数、Ｔ_ｅ２は予測モデルＢによる予測値である。

Here, t is time, T (t) is the measured temperature at time t, b is a coefficient, and _Te2 is a predicted value by the prediction model B.

By measuring the set of the measured temperature T (t) and the temperature change dT (t) / dt per unit time a plurality of times, the predicted value _Te2 and the coefficient b in the equation (4) can be estimated.

As shown in FIG. 1, the predicted value _Te2 by the prediction model B approaches the temperature T _pair to be measured over time as compared with the predicted value _Te1 by the prediction model A. However, the predicted value _Te2 by the prediction model B also has a large deviation from the temperature T _{pair of the} measurement target immediately after the temperature sensor is brought into contact with the measurement target.

As shown in FIG. 2 (a), the difference between the predicted value T _e2 by the measured temperature T _actual, prediction model B and the temperature T _versus the measured to the predicted value T _{e1, z-axis} by the prediction model A the y axis in the x-axis Is plotted to give a three-dimensional scatter plot. FIG. 2B is different from FIG. 2A in terms of viewpoint. As shown in FIGS. 2A and 2B, the plotted points are distributed in the vicinity of a plane having three dimensions. FIG. 2B is a three-dimensional scatter plot viewed from the viewpoint in this plane, and the plotted points are arranged substantially in a straight line. The parameters that specify this plane on three dimensions are α, β, and γ in the following equation (5). α, β, and γ are values peculiar to the temperature sensor.

ここで、Ｔ_対は測定対象の温度、Ｔ_実は実測温度、Ｔ_ｅ１は予測モデルＡによる予測値、Ｔ_ｅ２は予測モデルＢによる予測値、α、β、γは係数である。

Here, T _pairs of measured temperature, T _actual is measured temperature, T _e1 is predicted values by the prediction model A, T _e2 prediction value by the prediction model B, α, β, γ are coefficients.

In temperature prediction, so to predict is the predicted temperature of the measurement target, measured and equation (5) the temperature T _versus in the resulting deformed by replacing the predicted temperature T _pre, the following equation (6) Predicted temperature T _Used to predict the predictive temperature.

Measured temperature T _actual, since the predicted value T _e1 by prediction model _A, the predicted value T _e2 by prediction model B is available during the measurement, alpha relative to the temperature sensor, beta, that obtained in advance the γ the predicted temperature T _pre of the measuring object can be obtained from equation (6).

How to obtain α, β, and γ of the formula (5) will be described. An example of determining α and β after determining γ will be described below, but parameters α, β and γ may be obtained by principal component analysis or plane approximation.

As described above, α, β, and γ of the formula (5) are parameters for specifying the three-dimensional plane in which the points on the scatter plot of FIGS. 2 (a) and 2 (b) exist. When γ ≠ 0, when the scatter diagram of FIGS. 2 (a) and 2 (b) is projected onto a plane satisfying the predicted value _Te1 + (measured temperature T _actual / γ) = 0 by the prediction model A, the vertical axis is (predicted). A two-dimensional scatter diagram is obtained in which the predicted value _Te2 by the model B − the temperature T _{pair of the} measurement target) and the horizontal axis (predicted value _Te1 −γ by the predicted model A × the measured temperature T _actual ). In this embodiment, as shown in FIG. 3, a two-dimensional scatter plot with γ = 1 was obtained.

Also in the two-dimensional scatter plot shown in FIG. 3, it is judged that the plotted points are sufficiently on a straight line, and after determining γ = 1, (α, β) ≈ (1.1, 2. It was decided in 3). The correlation coefficient R ² was 0.9874.

Even in an experiment in which γ = 1 was fixed and the temperature T _{pair of the} measurement target was changed, the dependence of the parameters α and β on the temperature of the measurement target was low, and this set of parameters sufficiently represented the characteristics of the temperature sensor itself. It turned out to be a thing.

Determining a predicted temperature T _pre of the measurement target from the equation (6) as described above, but each of the predicted value T _e1, T _e2 of formula (3) and (4), when the temperature change per unit time is small Is susceptible to temperature measurement errors. Therefore, when the amount of temperature change is a predetermined value (for example, 0.08 ° C./sec) or less, it is desirable to avoid the prediction application.

Figure 4 shows actual temperature T _actual by the temperature sensor, the predicted temperature T _pre of the measuring object, the temperature T _pairs to be measured, a graph comparing the predicted value T _e1 by the prediction model A.

Wherein the predicted temperature T _pre of the measuring object obtained from (6) has a smaller deviation between the temperature T _pairs to be measured, immediately contacting the temperature sensor to the measurement object (for example, after lapse of 15 seconds) But the measured It can be seen that the deviation from the temperature T _pair is small.

Obtains the predicted value T _e1 by the prediction model A from equation (3), when obtaining the predicted value T _e2 by the prediction model B from equation (4), the temperature sensor to the temperature and measured temperature sensor prior to contact with the measurement object Does not require information on the time of contact. The same applies when obtaining the predicted temperature T _pre of the measurement target from the equation (6). Therefore, the temperature can be measured and the temperature can be predicted at an arbitrary time after the temperature sensor is installed on the measurement target.

Even when the predictable temperature range is wide from low temperature to high temperature and the temperature T _{pair to} be measured is lower than the temperature sensor before contact, it is predictable as shown in FIG.

Can temperature sensor to predict the predicted temperature T _pre of quickly measured than reach temperatures T _versus measured, can be terminated in a shorter time temperature prediction. Since the time required for temperature prediction is shortened, simultaneous measurement in combination with another quick-response contact sensor (for example, vibration sensor, acoustic emission sensor, strain sensor) is also possible.
(Temperature prediction method using multiple regression analysis)

Equation (5), that the measured temperature T _actual, relation of temperature T _versus predicted value _{T e2,} of the measurement target by the prediction value _{T e1,} prediction model B by the prediction model A, combined prediction model A, to B, the following It can also be obtained by multiple regression analysis as in.

ここで、Ｘ（１）は予測モデルＡによる予測値Ｔ_ｅ１であり、Ｘ（２）は実測温度である。□_ｍ（添え字のｍ）はそれぞれの平均値であり、□_ｓ（添え字のｓ）はそれぞれの標準偏差で標準化時に計算する。 X = (predicted value _Te1 by predicted model A1 _, measured temperature), Y = predicted value _Te2 by predicted model B-Temperature T _{pair to be} measured, and after standardization, the following equation (7) is estimated by multiple regression. To do. The coefficients c (1) and c (2) of the following equation (7) are estimated.

Here, X (1) is the predicted value _Te1 by the prediction model A, and X (2) is the measured temperature. □ _m (subscript m) is the average value of each, and □ _s (subscript s) is calculated at the time of standardization with each standard deviation.

By rearranging the equation (5), the following equation (8) can be obtained.

C (1) of the formula (8) corresponds to α of the formula (5), c (2) / c (1) of the formula (8) corresponds to γ of the formula (5), and the formula (8) The second term on the right side corresponds to β in equation (5). By estimating c (1) and c (2), α, β, and γ can be obtained.

6, alpha by multiple regression analysis, beta, and predicted temperature T _pre of the measuring object when called for gamma, alpha by a two-dimensional scatter diagram, beta, and predicted temperature T _pre of the measuring object when called for gamma It is a graph comparing. Substantially the same predicted temperature T _pre is obtained in either case.
(Temperature prediction method using prediction model C and prediction model D)

ここで、ｔは時間、Ｔ（ｔ）は時間ｔでの実測温度、ａは係数、Ｔ_ｅ３は予測モデルＣによる予測値である。 The prediction model A shown in the formula (3) was changed to the prediction model C shown in the following formula (9), and the prediction model B shown in the formula (4) was changed to the prediction model D shown in the following formula (10).

Here, t is time, T (t) is the measured temperature at time t, a is a coefficient, and _Te3 is a predicted value by the prediction model C.

ここで、ｔは時間、Ｔ（ｔ）は時間ｔでの実測温度、ｂは係数、Ｔ_ｅ４は予測モデルＤによる予測値である。

Here, t is time, T (t) is the measured temperature at time t, b is a coefficient, and _Te4 is a predicted value by the prediction model D.

α, β, and γ were determined by the above-mentioned multiple regression analysis.
Figure 7 is a graph comparing the predicted temperature T _pre using a predictive model B and the prediction model A, the predicted temperature T _pre using a predictive model D and the prediction model C. Substantially the same predicted temperature T _pre any case were obtained.

ｍは１以上の整数である。
（予測モデルＡと予測モデルＢと予測モデルＣを用いた温度予測方法） From FIG. 7, it can be seen that the prediction models A, B, C, and D can be predicted even if they are generalized as in the following equation (2).

m is an integer of 1 or more.
(Temperature prediction method using prediction model A, prediction model B, and prediction model C)

It predicted the predicted temperature T _pre of the measurement object by using a predictive model C shown in predictive models B and formula (9) shown in the prediction model A and Formula (4) shown in equation (3).

Ｔ_予は予測温度、Ｔ_ｅｉはｉ番目の予測モデルによる予測値、Ｔ_実は実測温度、α_ｉ、α_実、βは係数である。 Here, the relational expression (6) is generalized to the following equation (1) with respect to the number n of the prediction models.

T _pre is predicted _{temperature, T ei} is the prediction value by i-th prediction model, T _actual is measured temperature, alpha _i, alpha _real, beta is a coefficient.

Assuming that the model type is n = 2,

When the coefficient α ₂ is fixed at α ₂ = 1,

When the coefficient γ = α _real / α ₁ is introduced and replaced with α _real ,

Organize this

Equation (14) is the same as Equation (6) (however, the sign before the coefficient is reversed).

Figure 8 is a graph comparing the predicted temperature T _pre using predictive models A, B, predictive models A, B, and predicted temperature T _pre using C. Increasing the number of prediction models to three reduced the error immediately after the start of prediction. It seems that the prediction performance will improve by increasing the number of prediction models.
(Temperature prediction method using one type of prediction model)

It predicted the predicted temperature T _pre of the measurement object by using one type of prediction model A shown in Equation (3). As the relational expression, the expression (1) in which n = 1 was used.

Furthermore, it predicted predicted temperature T _pre of the measurement object by using one type of prediction model D shown in equation (10). As the relational expression, the expression (1) in which n = 1 was used.

9, compared predicted temperature T _pre using one type of prediction model A, one of the predicted temperature T _pre using a predictive model D, 2 kinds of prediction model A, the predicted temperature T _pre with B It is a graph. FIG. 10 is an enlarged graph of FIG. Temperature prediction was also possible using one type of prediction model A or one type of prediction model D. However, when two or more types of prediction models were used, the prediction performance was further improved.
(Comparative example not including the measured temperature in the relational expression)

In the formula (6), measured temperature T _fruit was examined whether mandatory. In the formula (6), the term of the _actual measured temperature T is simply deleted. The prediction model A, the predicted temperature T _pre with B (multiple regression analysis, invention example has measured temperature), predictive model C, the predicted temperature T _pre (multiple regression analysis using the D, the invention example has measured temperature ), prediction model a, the predicted temperature T _pre (multiple regression analysis using the B, Comparative example without actually measured temperature), predictive model C, the predicted temperature T _pre (multiple regression analysis using the D, Comparative examples without actually measured temperature ) Was compared.

Figure 11 is a graph comparing these predicted temperature T _pre. It was found that the accuracy was not correct when the measured temperature was not included in the relational expression.
<Temperature prediction device>

FIG. 12 is a block diagram showing an internal configuration of the temperature prediction device. The temperature prediction device 1 includes a temperature sensor 4, a processing unit 5, and a display device 2a. The temperature sensor 4 measures the temperature of the measurement target and outputs the temperature data to the processing unit 5. The temperature data output by the temperature sensor 4 is the measured temperature.

The processing unit 5 includes a CPU 5a and a memory 5b. The memory 5b includes a ROM for storing a temperature prediction program and a RAM for arithmetic processing. The CPU 5a operates according to the temperature prediction program. The CPU 5a realizes a predicted value calculating means and a temperature predicting means.

The temperature prediction device 1 of the present embodiment is portable and is used for a worker to attach a temperature sensor 4 to a bearing while patrolling the factory and measure the temperature of the bearing. A vibration sensor may be incorporated in the temperature sensor 4. In this case, the display device 2a displays the measurement results of the bearing temperature and vibration. Since it takes more time to measure the temperature than the measurement of the vibration, if the temperature prediction device 1 of the present embodiment is used, the measurement of the temperature can be completed at the same time as the measurement of the vibration. The measured temperature and vibration data may be transmitted to a personal computer or the like via the Internet. Of course, the uses and configurations of the temperature prediction device of the present invention are not limited to the above embodiments.

FIG. 13 shows a flowchart of the process executed by the temperature prediction device 1. This process is executed by the CPU 5a of the temperature prediction device 1. First, in S1, the power switch is turned on. After that, in S2, the temperature of the measurement target is measured by the temperature sensor.

In S3, it is determined whether or not the amount of temperature change is equal to or less than a predetermined value (for example, 0.08 ° C./sec) based on the measured temperature of the measurement target. If it is less than the predetermined value, it is easily affected by the temperature measurement error. Therefore, in S4, the measured temperature is displayed together with the message. If the amount of temperature change is equal to or greater than a predetermined value, the process proceeds to S5.

In S5, using the prediction model of n type (n ≧ 1), the predicted value _{T e1, T} e2 by the respective prediction model to calculate a ... _{T en.}

In S6, using equation (1), the measured temperature T _actual, predicted value by n different prediction model _{T e1, T} e2, obtaining the predicted temperature T _pre of ... _{T en.} In S7, the display of predicted temperature T _pre determined on the display unit 2a.
<Other embodiments>

In the above embodiment, an example in which the temperature prediction device is used as a device for predicting the temperature of the bearing has been described, but the temperature prediction device may be used as an electronic thermometer. In this case, the temperature predictor is formed so that it can be sandwiched by a person.

In the above embodiment, the example of executing the temperature prediction program by the temperature prediction device has been described, but the temperature prediction program may be executed by another device having a processor, for example, a computer such as a personal computer. In this case, the method of supplying the temperature prediction program is not limited, and for example, the temperature prediction program stored in a recording medium such as a CD-ROM may be installed in the computer, or the temperature may be installed from the server via the Internet line. You may download the prediction program.

1 ... Temperature prediction device 4 ... Temperature sensor 5a ... CPU (Prediction value calculation means, temperature prediction means)
S5 ... Predicted value calculation step S6 ... Temperature prediction step

Claims (7)

It is a temperature prediction method that predicts the temperature (T _pair ) of the measurement target based on the measured temperature (T _actual ) of the temperature sensor in contact with the measurement target.
Prediction by each prediction model using n types (n ≧ 1) of prediction models based on the temperature change of the measured temperature (T _real ) and the measured temperature (T _real ) of the measurement target by the temperature sensor per unit time. to calculate the value _{(T e1, T e2, ...} T en),
Previously obtained actual temperature and (T _actual) predicted value _{(T e1, T e2, ...} T en) using the predicted temperature (T _pre) and the relationship of the measured, temperature sensors measured temperature (T temperature predicting method for predicting the temperature of the measurement target (T _pair) as predicted temperature (T _pre) before _{the real)} matches the temperature (T _pairs) to be measured. The temperature prediction method according to claim 1, wherein the relational expression is represented by the following expression (1).

Here, T _pre is predicted temperature, T _ei is the prediction value by i-th prediction model, T _actual is measured temperature, alpha _i, alpha _real, beta is a coefficient. The temperature prediction method according to claim 1 or 2, wherein the following formula (2) is used for the prediction model.

Here, t is time, T (t) is the measured temperature at time t, T _e is the prediction value by the prediction model, m is an integer of at least 1, a is a coefficient. The temperature prediction method according to any one of claims 1 to 3, wherein n ≧ 2. Of the n types (n ≧ 2) of the prediction models, claim 4 is characterized in that the following formula (3) is used for the prediction model A and the following formula (4) is used for the prediction model B. The temperature prediction method described.

Here, t is time, T (t) is the measured temperature at time t, _Te1 is the predicted value by the prediction model A, _Te2 is the predicted value by the prediction model B, and a and b are coefficients. It is a temperature prediction device that predicts the temperature (T _pair ) of the measurement target based on the measured temperature (T _actual ) of the temperature sensor in contact with the measurement target.
Prediction by each prediction model using n types (n ≧ 1) of prediction models based on the temperature change per unit time of the measured temperature (T _real ) and the measured temperature (T _real ) of the measurement target by the temperature sensor. and the predicted value calculation means for calculating a value _{(T e1, T e2, ...} T en),
Previously obtained actual temperature and (T _actual) predicted value _{(T e1, T e2, ...} T en) using the predicted temperature (T _pre) and the relationship of the measured, temperature sensors measured temperature (T _real) temperature predicting apparatus and a temperature estimation means for estimating a predicted temperature (T _pre) the temperature of the measurement target (T _pairs) before matching the temperature (T _pairs) to be measured. It is a temperature prediction program that predicts the temperature (T _pair ) of the measurement target based on the measured temperature (T _actual ) of the temperature sensor in contact with the measurement target.
For devices with processors
Prediction by each prediction model using n types (n ≧ 1) of prediction models based on the temperature change per unit time of the measured temperature (T _real ) and the measured temperature (T _real ) of the measurement target by the temperature sensor. a prediction value calculating step of calculating a value _{(T e1, T e2, ...} T en),
Previously obtained actual temperature and (T _actual) predicted value _{(T e1, T e2, ...} T en) using the predicted temperature (T _pre) and the relationship of the measured, temperature sensors measured temperature (T _real) temperature predicting program for executing a temperature predicting step, the predicting the temperature of the measurement target (T _pairs) before matching the temperature (T _pairs) to be measured as a predicted temperature (T _pre).

Images