WO2019059000A1 - 雰囲気温度推定装置、雰囲気温度推定方法、プログラム及びシステム - Google Patents

雰囲気温度推定装置、雰囲気温度推定方法、プログラム及びシステム Download PDF

Info

Publication number
WO2019059000A1
WO2019059000A1 PCT/JP2018/033233 JP2018033233W WO2019059000A1 WO 2019059000 A1 WO2019059000 A1 WO 2019059000A1 JP 2018033233 W JP2018033233 W JP 2018033233W WO 2019059000 A1 WO2019059000 A1 WO 2019059000A1
Authority
WO
WIPO (PCT)
Prior art keywords
temperature
neural network
temperature estimation
value
estimation device
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/JP2018/033233
Other languages
English (en)
French (fr)
Japanese (ja)
Inventor
裕介 伴場
護 大柿
宏一 田野入
恵太 橋
卓也 松田
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Eizo Corp
Original Assignee
Eizo Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Eizo Corp filed Critical Eizo Corp
Priority to DE112018004283.9T priority Critical patent/DE112018004283T5/de
Priority to US16/649,351 priority patent/US11287860B2/en
Publication of WO2019059000A1 publication Critical patent/WO2019059000A1/ja
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01K—MEASURING TEMPERATURE; MEASURING QUANTITY OF HEAT; THERMALLY-SENSITIVE ELEMENTS NOT OTHERWISE PROVIDED FOR
    • G01K7/00—Measuring temperature based on the use of electric or magnetic elements directly sensitive to heat ; Power supply therefor, e.g. using thermoelectric elements
    • G01K7/42—Circuits effecting compensation of thermal inertia; Circuits for predicting the stationary value of a temperature
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
    • G06F1/16—Constructional details or arrangements
    • G06F1/20—Cooling means
    • G06F1/206—Cooling means comprising thermal management
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01K—MEASURING TEMPERATURE; MEASURING QUANTITY OF HEAT; THERMALLY-SENSITIVE ELEMENTS NOT OTHERWISE PROVIDED FOR
    • G01K1/00—Details of thermometers not specially adapted for particular types of thermometer
    • G01K1/20—Compensating for effects of temperature changes other than those to be measured, e.g. changes in ambient temperature
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D23/00—Control of temperature
    • G05D23/19—Control of temperature characterised by the use of electric means
    • G05D23/1917—Control of temperature characterised by the use of electric means using digital means
    • H—ELECTRICITY
    • H05—ELECTRIC TECHNIQUES NOT OTHERWISE PROVIDED FOR
    • H05K—PRINTED CIRCUITS; CASINGS OR CONSTRUCTIONAL DETAILS OF ELECTRIC APPARATUS; MANUFACTURE OF ASSEMBLAGES OF ELECTRICAL COMPONENTS
    • H05K7/00—Constructional details common to different types of electric apparatus
    • H05K7/20—Modifications to facilitate cooling, ventilating, or heating
    • H05K7/20954—Modifications to facilitate cooling, ventilating, or heating for display panels

Definitions

  • the present invention relates to an atmosphere temperature estimation device, an atmosphere temperature estimation method, a program, and a system.
  • Patent Document 1 discloses a display device capable of estimating the ambient temperature by using two or more temperature sensors in the display device and a correlation between temperatures detected by the temperature sensors.
  • Patent Document 1 enormous man-hours are required for positioning of a temperature sensor and adjustment of parameters used to estimate the ambient temperature. Specifically, in order to improve the estimation accuracy, it is necessary to manually adjust the parameters, which causes an increase in the number of steps. In addition, if the estimation accuracy can not be improved, positioning of the temperature sensor needs to be performed again, and a return occurs. Further, in Patent Document 1, it is a premise that the correlation between temperatures is determined under the condition where the cooling fan of the display device is driven under a constant control value. This is because, when the control value of the cooling fan is dynamically changed, the temperatures detected by two or more temperature sensors are also dynamically changed, and the correlation between the temperatures based on FIG. Because it is extremely difficult to do. Therefore, when the control value of the cooling fan in the display device is dynamically changed, in the technique of Patent Document 1, the estimation accuracy is reduced.
  • the present invention has been made in view of such circumstances, and an atmosphere temperature estimation device and atmosphere temperature estimation method capable of estimating atmosphere temperature with short man-hours and high accuracy without depending largely on the position of a temperature sensor. And provide a program. Further, even when the control value of the cooling fan is dynamically changed, an atmosphere temperature estimation device with high estimation accuracy of the atmosphere temperature, an atmosphere temperature estimation method, and a program are provided.
  • the ambient temperature around the atmosphere temperature estimation device is estimated using a neural network, a temperature acquisition unit that acquires one or more temperature values in the atmosphere temperature estimation device, and the neural network.
  • a neural network calculation unit is provided, and the input value input to the neural network by the neural network calculation unit is a temperature value acquired by the temperature acquisition unit, and a heat source control value for controlling a heat source in the atmosphere temperature estimation apparatus.
  • An ambient temperature estimation device is provided, including:
  • the atmosphere temperature is estimated by inputting parameters including one or more temperature values in the atmosphere temperature estimation device and a heat source control value for controlling the heat source in the atmosphere temperature estimation device to the neural network. . This made the estimation accuracy of the ambient temperature extremely high.
  • a cooling control unit that controls the inside of the atmosphere temperature estimation device based on a cooling control value
  • the input value includes the cooling control value.
  • the cooling control unit is configured to dynamically control the cooling control value.
  • the input value includes a change amount of the temperature value in a predetermined period.
  • the heat source is a backlight or an internal circuit.
  • the input value includes an energization time of at least one of the atmosphere temperature estimation device or the heat source.
  • the neural network is configured by a plurality of calculation nodes, and a predetermined weight is set for each of the calculation nodes, and the weight is set by machine learning in advance by another information processing apparatus. Or set by performing machine learning by the atmosphere temperature estimation apparatus.
  • the temperature acquisition unit acquires one or more temperature values in the atmosphere temperature estimation device, and the neural network calculation unit uses a neural network to obtain the atmosphere temperature estimation device.
  • a neural network calculation step of estimating the ambient temperature of the surroundings, and an input value inputted to the neural network controls a temperature value acquired by the temperature acquisition unit and a heat source in the atmosphere temperature estimation apparatus
  • An ambient temperature estimation method is provided that includes a heat source control value.
  • a neural network using a computer, a neural network, a temperature acquisition unit for acquiring one or more temperature values in the atmosphere temperature estimation device, and the neural network, the ambient temperature around the atmosphere temperature estimation device is It functions as a neural network calculation unit to estimate, and an input value input to the neural network by the neural network calculation unit controls a temperature value acquired by the temperature acquisition unit and a heat source in the atmosphere temperature estimation apparatus.
  • a program is provided that includes a heat source control value.
  • the atmosphere temperature estimation device according to any one of the above and the information processing device are included, and the atmosphere temperature estimation device and the information processing device are capable of data communication with each other.
  • the ambient temperature estimation device is configured to acquire the communication unit of the determined weights, a system is provided.
  • FIG. 2 is a conceptual diagram for describing a configuration of a neural network 20 and neural network calculation by a neural network calculator 14; It is a figure for demonstrating the weight w utilized for neural network calculation.
  • FIG. 4A is an example of a table representing the relationship between various parameters input to the neural network 20 and input signals
  • FIG. 4B is a relationship representing the relationship between an output signal output from the neural network 20 and the ambient temperature.
  • FIG. 2 is a functional block diagram of an information processing device 30 for determining weights w used for neural network calculation in the atmosphere temperature estimation device 1.
  • FIG. 6 is a conceptual diagram showing a state of machine learning using a neural network 50 of the information processing device 30.
  • FIG. 10A is a graph showing temporal changes in temperature values by two temperature sensors when the cooling fan is driven in a fixed manner
  • FIG. 10B is graphs showing temporal changes in temperature values by two temperature sensors when the cooling fan is variably controlled. is there. It is a graph which shows the estimation result which estimated atmosphere temperature using two temperature sensors, without applying this invention, and the estimation result of two types of atmosphere temperature by the atmosphere temperature estimation apparatus 1 of one Embodiment of this invention.
  • FIG. 10A is a graph showing temporal changes in temperature values by two temperature sensors when the cooling fan is driven in a fixed manner
  • FIG. 10B is graphs showing temporal changes in temperature values by two temperature sensors when the cooling fan is variably controlled. is there. It is a graph which shows the estimation result which estimated atmosphere temperature using two temperature sensors, without applying this invention, and the estimation result of two types of atmosphere temperature by the atmosphere temperature estimation apparatus 1 of one Embodiment of this invention.
  • FIG. 10A is a graph showing temporal changes in temperature values by two temperature sensors when the cooling fan is
  • FIG. 5 is a schematic view showing a relationship between an input gradation to the atmosphere temperature estimation device 1 and the luminance of the display unit 3;
  • FIG. 13A shows display characteristics by the atmosphere temperature estimation device 1 according to an embodiment of the present invention.
  • FIG. 13B shows display characteristics of a display device that does not estimate the ambient temperature.
  • FIGS. 1 to 7 An atmosphere temperature estimation apparatus 1 according to a first embodiment of the present invention will be described using FIGS. 1 to 7.
  • the atmospheric temperature estimation device 1 in the first embodiment can be applied to, for example, a display device.
  • the example which applied the atmospheric temperature estimation apparatus 1 to the display apparatus is demonstrated.
  • the atmosphere temperature estimation device 1 includes an operation unit 2, a display unit 3, a backlight 4, a cooling fan 5, a substrate 6, a temperature acquisition unit 7, a storage unit 8, a communication unit 9, a communication unit 9, a control unit 10 and A neural network 20 (represented as NN in the figure, and so forth) is provided.
  • the operation unit 2 operates the display unit 3 and includes, for example, a touch panel, a keyboard, a switch, a voice input unit, or a motion detection unit. For example, various setting information on the OSD (On Screen Display) is operated by the operation unit 2.
  • the display unit 3 displays various images (including a still image and a moving image), and includes, for example, a liquid crystal display, an organic EL display, an arbitrary touch panel display, and other displays.
  • the backlight 4 illuminates the display unit 3, and can be a heat source for releasing heat into the atmosphere temperature estimation device 1. Further, the backlight 4 is configured such that the intensity thereof can be controlled by a heat source control value described later.
  • the cooling fan 5 is provided in the atmosphere temperature estimation device 1 and cools the atmosphere temperature estimation device 1.
  • the cooling fan 5 is configured such that the drive strength thereof can be controlled by a cooling control value described later.
  • the substrate 6 is provided with various internal circuits provided in the ambient temperature estimation apparatus 1.
  • the internal circuit provided on the substrate 6 is an example of a heat source as in the backlight 4.
  • the heat source in the present embodiment is the backlight 4 or the internal circuit.
  • the atmosphere temperature estimation device 1 when the atmosphere temperature estimation device 1 is applied to a display device provided with an organic EL display, the organic EL display spontaneously emits light, so the backlight 4 is not an essential component.
  • the temperature acquisition unit 7 is configured of, for example, a temperature sensor, and is for acquiring one or more temperature values in the atmosphere temperature estimation device 1.
  • the temperature value may be an actual temperature itself, or may be a value subjected to predetermined conversion or a value proportional to or correlated with the actual temperature.
  • the temperature value when the temperature is 0 ° C. may be 0, and the temperature value when the temperature is 100 ° C. may be 1.
  • the temperature acquisition unit 7 can be provided at an arbitrary position in the atmosphere temperature estimation device 1.
  • the storage unit 8 stores various data and programs, and is configured of, for example, a memory, an HDD, an SSD, or the like.
  • the communication unit 9 transmits / receives various data to / from the control unit 10 or another information processing apparatus, and is configured by an arbitrary I / O.
  • the control unit 10 includes a heat source control unit 11, a cooling control unit 12, a temperature change amount calculation unit 13, and a neural network calculation unit 14 (displayed as an NN calculation unit in the figure; the same applies hereinafter).
  • the heat source control unit 11 controls the heat source in the atmosphere temperature estimation device 1 based on the heat source control value.
  • the backlight 4 is treated as a heat source
  • the heat source control value is a brightness setting value of the backlight 4.
  • the value used for the heat source control value is not particularly limited, but for example 0 (the preset value at which the brightness of the backlight 4 is the lowest (0)) to 100 (setting value at which the brightness of the backlight 4 is the maximum brightness) Can be
  • the cooling control unit 12 controls the inside of the ambient temperature estimation device 1 based on a cooling control value.
  • the cooling control value is a control value of the cooling fan 5.
  • the value used for the cooling control value is not particularly limited. For example, 0 (a set value at which the driving of the cooling fan 5 is at a minimum (stopping the cooling fan 5)) to 100 (a set value at which the driving of the cooling fan 5 is at maximum) can do.
  • the cooling control unit 12 is configured to dynamically control the cooling control. This makes it possible to dynamically change the drive strength of the cooling fan 5. Therefore, the driving strength of the cooling fan 5 is increased when the temperature in the atmosphere temperature estimation device 1 is high, while the driving strength of the cooling fan 5 is decreased when the temperature in the atmosphere temperature estimation device 1 is low. It is possible to reduce power consumption.
  • the temperature change amount calculation unit 13 calculates the change amount of the temperature value in a predetermined period from the temperature value acquired by the temperature acquisition unit 7.
  • the predetermined period is not particularly limited, and may be, for example, 1 second, 10 seconds, 30 seconds, 1 minute, 3 minutes, 5 minutes, 10 minutes or more or less.
  • the neural network calculation unit 14 estimates the ambient temperature around the ambient temperature estimation apparatus 1 using a neural network 20 described later.
  • the neural network calculation unit 14 inputs an input value based on various parameters to the neural network 20.
  • the input values input to the neural network 20 by the neural network calculation unit 14 include the temperature value acquired by the temperature acquisition unit 7 and the heat source control value for controlling the heat source in the atmosphere temperature estimation device 1.
  • the ambient temperature is a temperature that is not affected by the temperature rise due to the heat generation of the ambient temperature estimation device 1 or the like. That is, when the atmospheric temperature estimation device 1 is indoors, the temperature is substantially equal to room temperature, and when the atmospheric temperature estimation device 1 is outdoor, the temperature is substantially equal to the outdoor temperature.
  • the neural network calculation unit 14 inputs the temperature value acquired by the temperature acquisition unit 7 and the heat source control value for controlling the heat source in the atmosphere temperature estimation device 1 to the neural network 20, and the atmosphere temperature estimation device Output the ambient temperature around 1. Details of the neural network calculation unit 14 will be described later with reference to FIGS. 3 and 4.
  • the neural network 20 receives input values based on various parameters, is composed of a plurality of computing nodes N, and has a predetermined weight w set for each computing node N.
  • the weight w is a quantity that represents the coupling efficiency of the computation node N, and is also called coupling weight.
  • the neural network 20 can be implemented as software or hardware, for example, can be implemented on the firmware of the ambient temperature estimation device 1.
  • the weight w is set by machine learning in advance by another information processing apparatus 30 described later, or is set by machine learning performed in the atmosphere temperature estimation apparatus 1. Ru. Details will be described later.
  • the neural network 20 is composed of a plurality of layers (first to third layers L1 to L3) and a plurality of calculation nodes N (N11 to N31).
  • Nij represents the j-th computation node N in the i-th layer.
  • a predetermined weight w is set to each calculation node N. As shown in FIG. 3, for example, when focusing on the calculation node N23 of the second layer, a weight w is between the calculation node N23 and all the calculation nodes N11 to N15 of the first layer which is the immediately preceding layer. It is set.
  • the weight w is set to, for example, a value of -1 to 1.
  • the neural network calculation unit 14 inputs various parameters to the neural network 20.
  • a first temperature value, a second temperature value, a heat source control value, a cooling control value, a first temperature change amount, and a second temperature change amount are used as parameters to be input to the neural network 20.
  • the first temperature value and the second temperature value are temperature values based on the first temperature and the second temperature detected by the respective temperature sensors when two temperature sensors are provided as the temperature acquisition unit 7. .
  • the first temperature change amount and the second temperature change amount are change amounts in a predetermined period of time of the first temperature value and the second temperature value.
  • the input value input to the neural network 20 includes, in addition to the temperature value and the heat source control value, the cooling control value and the variation of the degree value in a predetermined period.
  • each parameter is normalized to a value of 0 to 1 when being input to the neural network 20.
  • all parameters are defined as 0 to 100, and a case where this is converted into an input signal of 0 to 1 will be described.
  • 0 first temperature value
  • 1 first temperature value
  • the first temperature is 100 ° C.
  • 1 heat source control value
  • the luminance setting value is 100
  • 1 heat source control value
  • the neural network calculation unit 14 inputs an input signal defined by various parameters to the first layer L1.
  • the input signals are output from the calculation nodes N11 to N15 in the first layer to the calculation nodes N21 to N25 in the second layer L2.
  • values obtained by multiplying the values output from the calculation nodes N11 to N15 by the weight w set for each calculation node N are input to the calculation nodes N21 to N25.
  • the calculation nodes N21 to N25 add together the input values, and input the value obtained by adding the bias b shown in FIG. 3 to the value to the activation function f ().
  • the output value of the activation function f () (the output value from the virtual calculation node node N'23 in the example of FIG.
  • the calculation node N31 which is the next node.
  • a value obtained by multiplying the output value with the weight w set between the calculation nodes N21 to N25 and the calculation node N31 is input to the calculation node N31.
  • the calculation node N31 adds the input values and outputs the sum as an output signal.
  • the computation node N31 may add together the input values, input a value obtained by adding a bias to the total value to the activation function, and output the output value as an output signal.
  • the value of the output signal is adjusted to be a value of 0 to 1.
  • the neural network calculation part 14 outputs the value corresponding to the value of an output signal as presumed atmosphere temperature.
  • the control unit 10 adjusts the luminance, the chromaticity, or the unevenness of the display unit 3 based on the estimated ambient temperature. For example, the control unit 10 rewrites the LUT table (not shown) to correct the gradation characteristics of the display unit 3 based on the estimated atmosphere temperature (gamma correction), so that the brightness of the display unit 3 can be Adjust to fit.
  • the ambient temperature estimation device 1 uses various parameters as input values, and machine learning is performed in advance by another information processing device 30 described later, or machine learning is performed by the atmosphere temperature estimation device 1.
  • the ambient temperature can be estimated with high accuracy as compared with the conventional case.
  • the amount of change in the temperature value in a predetermined period is used as the input value to the neural network 20, but instead of the amount of change in the temperature value, the atmosphere temperature estimation device 1 or the heat source Alternatively, at least one of the energization times may be used as an input value to the neural network 20.
  • the atmosphere temperature estimation apparatus 1 may be configured to use, as an input value, the energization time in a state in which the backlight is not energized but is energized.
  • the information processing apparatus 30 includes a communication unit 31, a parameter acquisition unit 32, a storage unit 33, a control unit 40, and a neural network 50.
  • the information processing apparatus 30 is configured by, for example, a computer or a server.
  • the communication unit 31, the storage unit 33, and the neural network 50 have the same functions as the communication unit 9, the storage unit 8, and the neural network 20 of the atmosphere temperature estimation device 1, and thus the description thereof will be omitted.
  • the communication unit 9 and the communication unit 31 are configured to be able to transmit and receive data to each other.
  • the number of layers in the neural network 50 and the configuration of the calculation node N can be designed as appropriate by the developer.
  • the parameter acquisition unit 32 acquires various parameters input to the neural network 20 of the atmosphere temperature estimation device 1.
  • the information processing device 30 acquires various parameters from the communication unit 9 of the atmosphere temperature estimation device 1 via the communication unit 31.
  • the connection mode of the communication unit 9 and the communication unit 31 is not particularly limited, and wired or wireless may be used.
  • the control unit 40 includes a temperature change calculation unit 41, a neural network calculation unit 42, a machine learning execution unit 43, and a weight setting unit 44.
  • the temperature change calculation unit 41 and the neural network calculation unit 42 have the same functions as the temperature change amount calculation unit 13 and the neural network calculation unit 14 of the atmosphere temperature estimation device 1, and thus the description thereof will be omitted.
  • the machine learning execution unit 43 executes machine learning for estimating the atmosphere temperature by substituting various parameters acquired by the parameter acquisition unit 32 into the neural network 50 and repeating the calculation by the neural network calculation unit 42. It is. Then, as a result of machine learning by the machine learning execution unit 43, the weight w is determined.
  • the weight setting unit 44 sets, in the neural network 50, the weight w to be set in the atmosphere temperature estimation device 1 as a result of machine learning by the machine learning execution unit 43.
  • the machine learning execution unit 43 sets, for example, a weight w of -1 to 1 for each calculation node N constituting the neural network 50 having the same configuration as the neural network 20 shown in FIG. Do. At this time, in order to reduce the influence of the weight w, it is preferable that the absolute value of the weight w set first is small. Then, the six types of input value sets acquired by the information processing device 30 are input to the neural network 50.
  • the output signal from the neural network 50 and teacher data representing the measured value of the atmosphere temperature are compared, and if the difference between the output signal and the teacher data (hereinafter referred to as error) is equal to or more than a predetermined threshold, the weight w is Change and input six sets of input values to the neural network 50 again. At this time, the change of the weight w is performed by a known error propagation method or the like. By repeatedly executing such calculation (machine learning), the error between the output signal from the neural network 50 and the pre-given teacher data is minimized. At this time, the number of times of learning of machine learning is not particularly limited, and can be, for example, 1000 times to 20000 times. In addition, even if the error between the output signal and the teacher data given in advance is not minimized, the machine learning is ended when the error becomes equal to or less than a predetermined threshold or at an arbitrary timing of the developer. You may
  • the weight setting unit 44 sets the weight of each calculation node N at this time in the neural network 50. That is, in the present embodiment, the weight w is stored in a storage unit such as a memory provided on the neural network 50. Then, the weight w set by the weight setting unit 44 is transmitted to the atmosphere temperature estimation device 1 via the communication unit 31 and is used as the weight of each calculation node N of the neural network 20 of the atmosphere temperature estimation device 1. In the present embodiment, the weight w is stored in a storage unit such as a memory provided on the neural network 20.
  • the configuration of the neural network 20 of the atmosphere temperature estimation device 1 the same as the configuration of the neural network 50 of the information processing device 30, it is possible to use the weight w set by the weight setting unit 44 as it is Become.
  • the developer sets the weight w by the information processing device 30 and mounts the weight w on the atmosphere temperature estimation device 1 before selling the atmosphere temperature estimation device 1.
  • the setting procedure of the weight w mounted in the atmosphere temperature estimation apparatus 1 will be described.
  • the ambient temperature estimation device 1 acquires six types of input values shown in FIGS. 2 and 6 by the temperature acquisition unit 7 or the like. For example, when the ambient temperature around the ambient temperature estimation device 1 is 25 ° C., it is assumed that the six types of input values are as follows. First temperature value: 0.25 Second temperature value: 0.3 Heat source control value: 0.5 Cooling control value: 0.3 First temperature change amount: 0.01 Second change in temperature: 0.02 The atmosphere temperature estimation apparatus 1 stores the above input values in the storage unit 8 as one set.
  • the ambient temperature estimation apparatus 1 acquires a temperature change in a predetermined period.
  • the atmosphere temperature estimation device 1 and the information processing device 30 are connected, and the plurality of input value sets are transmitted from the atmosphere temperature estimation device 1 to the information processing device 30.
  • the connection between the atmosphere temperature estimation device 1 and the information processing device 30 may be performed before S11 or S11.
  • the six types of input values can be transmitted as they are to the information processing device 30, without being stored in the storage unit 8 of the atmosphere temperature estimation device 1.
  • the information processing device 30 receives the plurality of input value sets acquired by the atmosphere temperature estimation device 1 in S21.
  • the neural network calculation unit 42, the machine learning execution unit 43, and the neural network 50 cooperate to execute the above-described machine learning.
  • the neural network calculation unit 42, the machine learning execution unit 43, and the neural network 50 cooperate to execute the above-described machine learning.
  • six types of input value sets are input to the first layer L1, weighting operations are performed in each calculation node, and an output signal is output from the neural network 50.
  • the weight setting unit 44 sets the weight w determined by the above-described machine learning in the neural network 50.
  • the process of S24 is not essential.
  • the atmosphere temperature estimation device 1 receives the weight w transmitted from the information processing device 30 in S13.
  • the weight w is set to the neural network 20 having the same structure as the neural network 50 of the information processing device 30.
  • the neural network 20 may be implemented on firmware in advance.
  • the result (weight w) machine-learned by the information processing apparatus 30 can be set in the neural network 20 of the atmosphere temperature estimation apparatus 1, and the atmosphere temperature estimation apparatus 1 of An ambient temperature estimation device 1 capable of estimating the ambient ambient temperature is realized.
  • the neural network 20 and the neural network 50 are composed of a plurality of computing nodes N, A predetermined weight w is set for each calculation node N,
  • the information processing device 30 is configured to obtain the temperature value acquired by the temperature acquisition unit 7 of the atmosphere temperature estimation device 1 and the heat source control value for controlling the heat source in the atmosphere temperature estimation device 1 by the communication unit 31.
  • the calculation node N and the weight w perform machine learning of the temperature value obtained from the atmosphere temperature estimation device 1 by the communication unit 31 of the information processing device 30 and the heat source control value using the neural network 50 of the information processing device 30.
  • Determined by The atmosphere temperature estimation device 1 is configured to acquire the determined weight w by the communication unit 9; system.
  • a plurality of (two) temperature acquiring units 7 can be provided at an arbitrary position in the atmosphere temperature estimation device 1. This is because the optimal weight w can be set by machine learning.
  • trial and error are required for the positions where the plurality of (two) temperature acquiring units 7 are provided, and a large number of man-hours are required, but the man-hours can be significantly reduced in this embodiment.
  • a large number of man-hours are required to determine the correlation between the temperature values acquired by the two temperature acquiring units 7, but in the present embodiment, such man-hours are significantly reduced by using machine learning. be able to. Thereby, high robustness to the installation position of the temperature acquisition unit 7 and reduction of man-hours can be realized as compared with the conventional case.
  • the atmosphere temperature estimation apparatus 1 of the first embodiment exhibits the following effects. As compared with the prior art, the ambient temperature can be estimated with extremely high accuracy. Furthermore, since the developer determines the structure and weight w of the neural network 20, it is possible to reduce the decrease in estimation accuracy due to so-called "over-learning".
  • the atmosphere temperature estimation apparatus 1 of the second embodiment differs from the first embodiment in that a machine learning execution unit 15 and a weight setting unit 16 are provided. The differences will be described below.
  • the atmosphere temperature estimation apparatus 1 of the second embodiment includes the machine learning execution unit 15 and the weight setting unit 16, even after the atmosphere temperature estimation apparatus 1 reaches the user's hand, machine learning is performed by itself and the atmosphere temperature is The estimation accuracy of can be improved.
  • the weight w it is preferable to set in advance the weight w using the information processing device 30 at the time of shipment of the atmosphere temperature estimation device 1. Then, after the atmosphere temperature estimation device 1 is shipped, it is possible to improve the estimation accuracy of the atmosphere temperature of the atmosphere temperature estimation device 1 daily by updating the preset weight w itself.
  • a system 60 according to a third embodiment will be described with reference to FIG.
  • a plurality of atmosphere temperature estimation devices 1 are connected to the information processing device 30 via the network 100.
  • Each atmosphere temperature estimation device 1 transmits various parameters to be input to the neural network 20 to the information processing device 30 after being at the user's hand. Then, machine learning is performed by the information processing apparatus 30 operated by the developer, and the updated weight w is transmitted to the atmosphere temperature estimation apparatus 1.
  • the atmosphere temperature estimation device 1 can improve the estimation accuracy of the atmosphere temperature of the atmosphere temperature estimation device 1 by setting the updated weight w in each calculation node N of the neural network 20.
  • the weight w is updated by the developer even after the atmosphere temperature estimation device 1 is shipped, so that progress of inappropriate machine learning on the user side can be reduced. .
  • FIG. 10A and FIG. 10B are figures for demonstrating the meaning of the atmospheric temperature estimation apparatus 1 of the said embodiment.
  • FIG. 10A shows the temperature detected by the temperature acquisition unit 7 (first temperature sensor and second temperature sensor) in the atmosphere temperature estimation device 1 when the cooling control value for controlling the drive strength of the cooling fan 5 is held constant. It is a graph which shows a time-dependent change of a value.
  • FIG. 10B shows the temperature acquisition unit 7 (first temperature sensor and second temperature sensor) in the atmosphere temperature estimation device 1 when the cooling control value for controlling the drive strength of the cooling fan 5 is dynamically changed. It is a graph which shows a time-dependent change of the detected temperature value.
  • the temperature detected by the temperature sensor is normalized and indicated on the vertical axis as a temperature value (normalized).
  • an atmosphere temperature estimation device using the correlation between temperatures detected by two temperature sensors as in the technique of Patent Document 1, for example. It is possible to determine a relational expression that estimates the ambient temperature around 1.
  • FIG. 11 is a graph in which the actual ambient temperature (thermocouple measurement) and the estimated ambient temperature are plotted in the case of using the ambient temperature estimation apparatus 1 of the above embodiment.
  • the vertical axis represents the temperature
  • the horizontal axis represents the time after the brightness of the backlight 4 is reduced.
  • estimation by estimation (NN estimation) using two types of neural networks 20 The temperature is plotted.
  • two types of NN estimation are estimated without using the temperature change amount calculated by the temperature change amount calculation unit 13 (NN estimation (no temperature change amount)) and the temperature change amount.
  • the estimation results obtained when the actual ambient temperature is 15 ° C, 25 ° C, and 35 ° C are collectively plotted.
  • the cooling fan 5 was stopped when the actual ambient temperature was 15.degree. C., and the cooling fan 5 was driven at a constant strength when the actual ambient temperature was 25.degree. C. and 35.degree. This is because when the ambient temperature is 15.degree. C., it is less necessary to cool the inside of the ambient temperature estimation device 1 by the cooling fan 5, and the actual usage scene is reproduced.
  • the ambient temperature is 15 ° C., that is, when the cooling fan 5 is stopped
  • the error between the conventional method and the actual ambient temperature is large.
  • Such an error is at most 4 ° C., and the estimation accuracy of the ambient temperature is low. It is considered that this is because the temperature in the atmosphere temperature estimation device 1 changes randomly (corresponding to the state of FIG. 10B) because the cooling fan 5 is stopped. It is also considered that the internal temperature has risen compared to when the fan is driven at a constant strength.
  • the estimation accuracy is further improved by using the temperature change amount.
  • FIGS. 12 and 13 are graphs showing the relationship between the input gradation and the luminance of the display unit 3 in the prior art and the embodiment.
  • the gamma curve (solid line 1) set in the atmosphere temperature estimation apparatus 1 and the measured luminance (dotted line 2) of the actual display unit 3
  • the measured luminance of the display unit 3 follows the solid line 1 If the estimation accuracy of the ambient temperature is low, as indicated by the dotted line 2, the measured luminance of the ambient temperature estimation device 1 deviates up and down from the ideal dotted line 1.
  • the deviation from the solid line 1 upward is defined as “error rate: +”
  • the deviation from the solid line 1 downward is defined as “error rate: ⁇ ”.
  • FIG. 13 is a graph showing the error rate of the luminance of the display unit 3 when the cooling control value of the cooling fan 5 is dynamically changed under a plurality of ambient temperatures.
  • the vertical axis represents the error rate
  • the horizontal axis represents the gradation value of RGB. Note that, in FIG. 13, all RGB are set to the same gradation value.
  • the plurality of atmospheric temperatures in FIG. 13 are the same as those in FIG. Specifically, the error rates under the following 12 conditions.
  • FIG. 13A shows the error rate of the brightness of the display unit 3 when the prior art is used
  • FIG. 13B shows the error rate of the brightness of the display unit 3 when the above embodiment is used.
  • a temperature acquisition unit for acquiring one or more temperature values in the atmosphere estimation device;
  • a neural network calculation unit for estimating the ambient temperature around the atmosphere estimation device using the neural network;
  • the input value input to the neural network by the neural network calculation unit includes the temperature value acquired by the temperature acquisition unit and a heat source control value for controlling a heat source in the atmosphere estimation device. program.
  • the cooling control value of the cooling fan 5 can be coped with by various changes including linear change. -It is also possible to use only one temperature acquisition unit 7.
  • the number of calculation nodes N and the number of layers can adopt arbitrary values.
  • substrate can be used as a heat source, and the control value with respect to this internal circuit or a board
  • the embodiment which estimates external atmosphere temperature only by the information inside a display apparatus as an example of the atmosphere temperature estimation apparatus 1 was described. This is because when the temperature sensor is provided outside the display device, it is difficult to accurately measure the ambient temperature because it is affected by the heat generation of the display device.
  • the present invention can also be applied to various devices having the same problem (devices that estimate the external ambient temperature only by the information in the device).
  • Atmosphere temperature estimation device 2 Operation unit 3: Display unit 4: Back light 5: Cooling fan 6: Substrate 7: Temperature acquisition unit 8: Storage unit 9: Communication unit 10: Control unit 11: Heat source control unit 12: Cooling Control unit 13: Temperature change amount calculation unit 14: Neural network calculation unit 20: Neural network 30: Information processing device 31: Communication unit 32: Parameter acquisition unit 33: Storage unit 40: Control unit 41: Temperature change calculation unit 42: Neural Network calculation unit 43: machine learning execution unit 44: weight setting unit 50: neural network 60: system

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Automation & Control Theory (AREA)
  • Human Computer Interaction (AREA)
  • Measuring Temperature Or Quantity Of Heat (AREA)
  • Control Of Temperature (AREA)
  • Software Systems (AREA)
  • Feedback Control In General (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
PCT/JP2018/033233 2017-09-25 2018-09-07 雰囲気温度推定装置、雰囲気温度推定方法、プログラム及びシステム Ceased WO2019059000A1 (ja)

Priority Applications (2)

Application Number Priority Date Filing Date Title
DE112018004283.9T DE112018004283T5 (de) 2017-09-25 2018-09-07 Vorrichtung zum schätzen einer umgebungstemperatur, verfahren zum schätzen einer umgebungstemperatur, programm und system
US16/649,351 US11287860B2 (en) 2017-09-25 2018-09-07 Ambient temperature estimating device, ambient temperature estimating method, program and system

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2017183582A JP6723964B2 (ja) 2017-09-25 2017-09-25 雰囲気温度推定装置、雰囲気温度推定方法、プログラム及びシステム
JP2017-183582 2017-09-25

Publications (1)

Publication Number Publication Date
WO2019059000A1 true WO2019059000A1 (ja) 2019-03-28

Family

ID=65810857

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2018/033233 Ceased WO2019059000A1 (ja) 2017-09-25 2018-09-07 雰囲気温度推定装置、雰囲気温度推定方法、プログラム及びシステム

Country Status (4)

Country Link
US (1) US11287860B2 (enExample)
JP (1) JP6723964B2 (enExample)
DE (1) DE112018004283T5 (enExample)
WO (1) WO2019059000A1 (enExample)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2021157930A1 (en) * 2020-02-03 2021-08-12 Samsung Electronics Co., Ltd. Methods and systems for ascertaining factors contributing to the temperature of a device
JPWO2022009345A1 (enExample) * 2020-07-08 2022-01-13
JPWO2022054148A1 (enExample) * 2020-09-09 2022-03-17
CN117589334A (zh) * 2024-01-19 2024-02-23 湖南华夏特变股份有限公司 油浸式变压器热点温度检测方法及系统

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP7378063B2 (ja) * 2019-10-25 2023-11-13 国立大学法人秋田大学 温度推定装置及び温度推定方法
KR102862457B1 (ko) * 2020-08-11 2025-09-23 삼성전자주식회사 전자 장치 및 그 제어 방법
EP4323813A4 (en) * 2021-04-14 2025-01-22 Ecosystem Informatics Inc. MONITORING AN AMBIENT AIR PARAMETER USING A TRAINED MODEL
WO2022218542A1 (en) * 2021-04-16 2022-10-20 Nokia Solutions And Networks Oy An apparatus, a method, and a non-transitory computer readable medium for determining a random access preamble identifier
US20230392990A1 (en) * 2022-06-06 2023-12-07 Apple Inc. Internal heat-generating component noise correction techniques for temperature sensor systems
CN118265275B (zh) * 2024-05-30 2024-08-27 山西新泰富安新材有限公司 风冷线控冷优化控制方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002048878A (ja) * 2000-08-02 2002-02-15 Nagoya Electric Works Co Ltd 路面温度または路面状態の予測方法およびその装置
US20060165149A1 (en) * 2005-01-26 2006-07-27 Carrier Corporation Dynamic correction of sensed temperature
US20140355649A1 (en) * 2013-05-31 2014-12-04 Sensirion Ag Portable electronic device with integrated temperature sensor
JP2015534038A (ja) * 2012-07-02 2015-11-26 ゼンジリオン・アーゲー ポータブル電子デバイス

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6995355B2 (en) * 2003-06-23 2006-02-07 Advanced Optical Technologies, Llc Optical integrating chamber lighting using multiple color sources
JP4228931B2 (ja) * 2004-02-18 2009-02-25 日本電気株式会社 液晶パネルの駆動装置及び駆動方法並びにこれを用いた液晶プロジェクタ
EP2434474A4 (en) * 2009-05-19 2013-03-27 Sharp Kk Liquid crystal display device and method for its activation
JP5566780B2 (ja) 2010-06-01 2014-08-06 富士フイルム株式会社 赤外線用結像レンズおよび撮像装置
JP5080681B1 (ja) * 2011-11-18 2012-11-21 株式会社ナナオ 表示装置、コンピュータプログラム、記録媒体及び温度推定方法
JP2013134384A (ja) * 2011-12-27 2013-07-08 Seiko Epson Corp プロジェクター、及び、プロジェクターの制御方法
JP2019129489A (ja) * 2018-01-26 2019-08-01 セイコーエプソン株式会社 集積回路装置、振動デバイス、電子機器及び移動体

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002048878A (ja) * 2000-08-02 2002-02-15 Nagoya Electric Works Co Ltd 路面温度または路面状態の予測方法およびその装置
US20060165149A1 (en) * 2005-01-26 2006-07-27 Carrier Corporation Dynamic correction of sensed temperature
JP2015534038A (ja) * 2012-07-02 2015-11-26 ゼンジリオン・アーゲー ポータブル電子デバイス
US20140355649A1 (en) * 2013-05-31 2014-12-04 Sensirion Ag Portable electronic device with integrated temperature sensor

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2021157930A1 (en) * 2020-02-03 2021-08-12 Samsung Electronics Co., Ltd. Methods and systems for ascertaining factors contributing to the temperature of a device
US11971698B2 (en) 2020-02-03 2024-04-30 Samsung Electronics Co., Ltd. Methods and systems for ascertaining factors contributing to the temperature of a device
JPWO2022009345A1 (enExample) * 2020-07-08 2022-01-13
WO2022009345A1 (ja) * 2020-07-08 2022-01-13 三菱電機株式会社 画像処理装置、表示制御装置、画像表示装置、画像処理方法、プログラム、及び記録媒体
JP7471416B2 (ja) 2020-07-08 2024-04-19 三菱電機株式会社 画像処理装置、表示制御装置、画像表示装置、画像処理方法、プログラム、及び記録媒体
JPWO2022054148A1 (enExample) * 2020-09-09 2022-03-17
WO2022054148A1 (ja) * 2020-09-09 2022-03-17 三菱電機株式会社 画像処理装置、画像表示装置、画像処理方法、及び画像処理プログラム
JP7275402B2 (ja) 2020-09-09 2023-05-17 三菱電機株式会社 画像処理装置、画像表示装置、画像処理方法、及び画像処理プログラム
CN117589334A (zh) * 2024-01-19 2024-02-23 湖南华夏特变股份有限公司 油浸式变压器热点温度检测方法及系统
CN117589334B (zh) * 2024-01-19 2024-03-26 湖南华夏特变股份有限公司 油浸式变压器热点温度检测方法及系统

Also Published As

Publication number Publication date
US20200301487A1 (en) 2020-09-24
JP2019060642A (ja) 2019-04-18
DE112018004283T5 (de) 2020-06-25
US11287860B2 (en) 2022-03-29
JP6723964B2 (ja) 2020-07-15

Similar Documents

Publication Publication Date Title
JP6723964B2 (ja) 雰囲気温度推定装置、雰囲気温度推定方法、プログラム及びシステム
US20220130329A1 (en) System and methods for extracting correlation curves for an organic light emitting device
US11410614B2 (en) System and method for loading image correction data for displays
CN107452345A (zh) 一种显示器件亮度补偿方法、装置及显示装置
CN102741910A (zh) 用于提取有机发光器件的相关曲线的系统和方法
RU2613532C1 (ru) Многомониторная система и используемые в ней компьютерная программа и отображающее устройство
CN106409219B (zh) 调整面板亮度和色度的预补偿方法及装置
US9514699B2 (en) Device and method for adjusting gamma voltage
KR20160074853A (ko) 표시 장치 및 표시 장치의 구동 방법
CN109949750B (zh) 显示装置及其驱动方法
CN106448544A (zh) Led显示装置
CN112002285B (zh) 用于确定和补偿有机发光器件的效率劣化的方法
KR20160059021A (ko) 표시 패널 구동 장치 및 이를 포함하는 표시 장치
CN109686306A (zh) 补偿因子获取方法及装置、驱动方法、显示设备
KR20170003251A (ko) 유기 발광 다이오드 표시장치 및 구동방법
CN109147674B (zh) Amoled显示残影消除方法、显示终端及存储介质
CN105097872B (zh) 提取有机发光器件的关联曲线的系统和方法
CN110021268B (zh) Oled的显示控制方法和装置
CN110853578A (zh) 电压补偿方法和装置
KR102852267B1 (ko) 데이터 보상 회로, 이를 포함하는 표시 장치 및 이를 이용한 데이터 보상 방법
JP2017032890A (ja) Ｌｅｄ表示装置
JP2017032889A (ja) Ｌｅｄ表示装置
CN109410892B (zh) 影像处理方法及显示装置
KR20160028591A (ko) 평판 표시장치 및 그의 구동방법
US11804189B2 (en) Display device, method for generating offset current values and current offsetting system

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18858177

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 18858177

Country of ref document: EP

Kind code of ref document: A1