Industrial computer heat dissipation control method and system

Through the neural network model and Gaussian distribution model combined with ventilation evaluation, the speed of the industrial control fan is adjusted in advance, solving the problem of temperature adjustment hysteresis of the industrial control machine and improving the response speed and heat dissipation efficiency of temperature adjustment.

CN118548238BActive Publication Date: 2024-11-29SUZHOU NUODAJIA AUTOMATION TECH CO LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411016914.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-29
Publication Date
2024-11-29
Estimated Expiration
2044-07-29

AI Technical Summary

Technical Problem

The fan gear control of existing industrial control machines has a temperature adjustment hysteresis, which leads to the fan speed not adjusting in time when the temperature changes rapidly, resulting in waste of energy and additional noise.

Method used

The neural network model is used to combine Gaussian distribution model and ventilation evaluation. By obtaining the gradient value of the air outlet temperature image, the ventilation evaluation is calculated, and the cooling gear of the industrial control machine is adjusted according to the temperature prediction value, and the fan speed is adjusted in advance.

Benefits of technology

It reduces the hysteresis effect of temperature regulation, improves the response speed of fan speed, reduces energy waste and noise, and improves heat dissipation efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118548238B_ABST
    Figure CN118548238B_ABST
Patent Text Reader

Abstract

The present application relates to the field of heat dissipation control, and in particular to a heat dissipation control method and system for an industrial computer, the method comprising: obtaining temperature images of continuous frames at an air outlet, calculating the gradient values ​​of pixel points in the temperature image, and constructing a gradient value set according to the gradient values ​​of each pixel point; inputting the gradient value set into a preset Gaussian distribution model to generate a Gaussian distribution curve, and taking the pixel points corresponding to the gradient values ​​of the Gaussian distribution curve outside / inside a preset interval as heat dissipation points; calculating ventilation evaluation; inputting the ventilation evaluation and the real-time temperature into a preset prediction model to generate a temperature prediction value, and adjusting the heat dissipation gear of the industrial computer according to the temperature prediction value, wherein each heat dissipation gear corresponds to a prediction model. The present application adjusts the speed of the fan in advance according to the actual heat dissipation situation, and has the effect of reducing the hysteresis of temperature regulation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of heat dissipation control, and in particular to a heat dissipation control method and system for an industrial computer. Background Art

[0002] Industrial Personal Computer (IPC) is a computer designed for industrial environments. IPCs usually run for a long time in environments with high temperature, dust, and poor power supply conditions. Good heat dissipation design is essential to ensure its stability.

[0003] Industrial computers can be equipped with fans with multiple speed levels to achieve more precise temperature control. For example, when the load is low, the fan speed level can be low to reduce noise and energy consumption, while when the load is high, the fan speed can be increased to enhance the cooling effect.

[0004] In the prior art, for example, a Chinese patent with the authorization announcement number CN208207694U discloses a temperature-controlled industrial control chassis, including: a power adapter located at the right rear end of the chassis, a routing board located at the middle and rear end of the chassis, a first PCB board located at the middle and front end of the chassis, a second PCB board, a third PCB board and a fourth PCB board located at the left side of the chassis, an RJ45 port located at the rear end of the chassis, a banana pie and a hard disk located at the right side of the chassis, and a temperature control board and a cooling fan located at the front end of the chassis. The cooling fan is automatically adjusted and controlled by the temperature control board.

[0005] In the prior art, the gear control of the fan of the industrial computer is performed after the temperature rises or drops for a period of time, that is, there is a lag in the speed of responding to temperature changes during the temperature adjustment process. This lag may cause the fan speed to be adjusted inadequately when the temperature rises rapidly, or the fan may run at a high speed when it is not necessary, resulting in energy waste and additional noise. Summary of the invention

[0006] In order to reduce the hysteresis effect of temperature regulation and adjust the fan speed in advance according to the actual heat dissipation situation, the present application provides an industrial computer heat dissipation control method and system.

[0007] In a first aspect, the present application provides a method for controlling heat dissipation of an industrial computer, which adopts the following technical solution:

[0008] A heat dissipation control method for an industrial computer comprises the steps of: obtaining temperature images of continuous frames at an air outlet, calculating gradient values ​​of pixel points in the temperature image, and constructing a gradient value set according to the gradient values ​​of each pixel point; inputting the gradient value set into a preset Gaussian distribution model to generate a Gaussian distribution curve, and taking the pixel points corresponding to the gradient values ​​outside / inside the preset interval of the Gaussian distribution curve as heat dissipation points; and calculating ventilation evaluation, wherein the calculation formula of the ventilation evaluation is: ,in, Indicates ventilation rating, Indicates the number of heat dissipation points in the current frame temperature image. Indicates the number of heat dissipation points in the next frame temperature image of the current frame. Represents the set of pixel values ​​of heat dissipation points in the temperature image of the current frame, Represents the set of pixel values ​​of heat dissipation points in the temperature image of the next frame of the current frame, Represents the average; inputs the ventilation evaluation and the real-time temperature into the preset prediction model to generate a temperature prediction value, and adjusts the heat dissipation level of the industrial computer according to the temperature prediction value, wherein each heat dissipation level corresponds to a prediction model.

[0009] The beneficial effects are as follows: the present application combines a neural network model to predict temperature values, thereby achieving the purpose of predicting temperature changes, adjusting the heat dissipation gear according to the temperature prediction value, and being able to adjust the fan speed in advance according to the actual heat dissipation situation, thereby reducing the lag effect of temperature regulation. Among them, the ventilation evaluation is used to measure the heat dissipation efficiency or ventilation effect of the heat dissipation point between two frames of temperature images, and quantifies the change in ventilation effect by comparing the heat dissipation quantity and temperature average of the heat dissipation point in the two frames of images.

[0010] Optionally, the method for calculating the gradient value of a pixel point in a temperature image includes the steps of: obtaining the pixel value of any pixel point and the maximum value of the pixel values ​​of eight neighborhood pixels centered on the pixel point, and taking the ratio of the maximum value to the pixel value of the pixel point as the gradient value of the pixel point; traversing all pixel points to obtain the gradient values ​​of all pixel points.

[0011] The beneficial effect is that by comparing the pixel values ​​of the central pixel with those in its eight neighborhoods, the area with the largest local changes can be highlighted, which is helpful for feature recognition.

[0012] Optionally, the preset interval is Gaussian distribution ( ) interval, is the mean of the data in the Gaussian distribution, Represents the standard deviation of data from a Gaussian distribution.

[0013] Optionally, the cooling gear of the industrial computer is adjusted according to the temperature prediction value, including the steps of: in response to the real-time temperature average within the preset time being greater than the predicted temperature average, increasing the cooling gear to increase the fan speed; in response to the real-time temperature average within the preset time being less than the predicted temperature average, decreasing the cooling gear to reduce the fan speed; in response to the real-time temperature average within the preset time being equal to the predicted temperature average, maintaining the current cooling gear.

[0014] Optionally, set a data set, which includes historical temperature data of the industrial computer and the corresponding ventilation evaluation; divide the data set into a training set and a validation set, and input the training set into the neural network model for model training until the preset model training stop condition is reached to complete the training and obtain a prediction model.

[0015] Optionally, modify the loss function during training.

[0016] The beneficial effect is: when using the neural network model for prediction, the prediction results may be over-discrete. In order to reduce the number of training times, the loss function is corrected during the training process, and the introduction of the correction coefficient enables the loss function to be adaptively adjusted according to the deviation between the predicted value and the actual fitting curve, thereby enhancing the model's adaptability to different temperature change patterns.

[0017] Optionally, the loss function is corrected during the training process, including the steps of: using the least squares method to fit the temperature collection values ​​at multiple moments before the prediction moment with a spline curve to obtain a fitted nonlinear curve; calculating the distance between the temperature prediction value at the prediction moment and the fitted nonlinear curve, and obtaining a correction coefficient based on the distance. The expression of the correction coefficient is: ,in, Indicates The correction coefficient corresponding to the temperature prediction value is: Indicates The distance between the predicted temperature value and the fitted nonlinear curve, Represents the total number of temperature prediction values; the correction coefficient is used as the weight of the loss function to obtain the corrected loss function.

[0018] The beneficial effect is that by introducing spline curve fitting and distance calculation, the changing trend of temperature data can be captured more accurately, thereby improving the accuracy of temperature prediction.

[0019] In a second aspect, the present application provides an industrial computer heat dissipation control system, which adopts the following technical solution:

[0020] A heat dissipation control system for an industrial computer comprises: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the heat dissipation control method for the industrial computer is implemented.

[0021] The beneficial effect is that the above-mentioned heat dissipation control method for an industrial computer is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a system based on the memory and the processor for easy use.

[0022] This application has the following technical effects:

[0023] 1. This application combines a neural network model to predict temperature values, thereby achieving the purpose of predicting temperature changes. The heat dissipation gear is adjusted according to the temperature prediction value, and the fan speed can be adjusted in advance according to the actual heat dissipation situation, thereby reducing the lag effect of temperature regulation.

[0024] 2. By measuring the heat dissipation efficiency or ventilation effect of the heat dissipation point between two frames of temperature images, the heat dissipation quantity and the average temperature of the heat dissipation point in the two frames of images are compared to quantify the change in ventilation effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present application will become easily understood. In the accompanying drawings, several embodiments of the present application are shown in an exemplary and non-restrictive manner, and the same or corresponding numbers represent the same or corresponding parts.

[0026] Figure 1 This is a method flow chart of a heat dissipation control method for an industrial computer according to an embodiment of the present application.

[0027] Figure 2 This is a method flow chart of S1 in a method for controlling heat dissipation of an industrial computer in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by technicians in the field without creative work are within the scope of protection of the present application.

[0029] It should be understood that when the terms "first", "second", etc. are used in the claims, specification and drawings of the present application, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the specification and claims of the present application indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their collections.

[0030] The present application embodiment discloses a method for controlling heat dissipation of an industrial computer, referring to Figure 1 , including steps S1 to S4, which are specifically as follows:

[0031] S1: Obtain temperature images of continuous frames at the air outlet, calculate the gradient values ​​of pixels in the temperature image, and construct a gradient value set according to the gradient values ​​of each pixel.

[0032] A thermal imager is set at the external air outlet of the industrial computer to collect the temperature image of the external air outlet of the industrial computer. The temperature image refers to a thermal imaging image or a thermal image. A temperature sensor is set inside the industrial computer to collect the temperature changes of the electronic components inside the industrial computer.

[0033] For temperature images, the size of the pixel value can reflect the temperature. The larger the pixel value, the higher the temperature of the pixel point, and the smaller the pixel value, the smaller the temperature of the pixel point.

[0034] Reference Figure 2 The method for calculating the gradient value of a pixel point in a temperature image includes steps S10-S11, which are as follows:

[0035] S10: Obtain a pixel value of any pixel point and the maximum value of pixel values ​​of eight neighboring pixels centered on the pixel point, and take the ratio of the maximum value to the pixel value of the pixel point as the gradient value of the pixel point.

[0036] Specifically, an eight-neighborhood system centered on each pixel is determined, where the eight-neighborhood refers to a neighborhood consisting of eight pixels around a pixel, including the neighboring pixels above, below, left, right, and four diagonal directions of the pixel.

[0037] Taking a pixel as an example, obtain the temperature values ​​of all pixels in the eight neighborhoods of the pixel, find the maximum temperature value in the eight neighborhoods, and take the ratio of the maximum value to the pixel value of the pixel as the gradient value of the pixel.

[0038] S11: Traverse all pixel points to obtain the gradient values ​​of all pixel points.

[0039] The above operation is performed on all pixel points to complete the gradient value calculation of all pixel points, and a gradient value set is constructed based on the calculated gradient values.

[0040] S2: Input the gradient value set into a preset Gaussian distribution model to generate a Gaussian distribution curve, and use the pixel points corresponding to the gradient values ​​outside / inside the preset interval of the Gaussian distribution curve as heat dissipation points.

[0041] The preset interval is Gaussian distribution ( ) interval, or ( ) interval, is the mean of the data in the Gaussian distribution, Represents the standard deviation of data in a Gaussian distribution. Gaussian distribution is also called normal distribution. The principle is: the values ​​are distributed in ( ) has a probability of 0.6827; the value distribution is ( ) has a probability of 0.9545; the value distribution is in ( ) has a probability of 0.9973. It can be considered that the values ​​are almost all concentrated in ( ) range, and the probability of exceeding this range is about 0.3%.

[0042] In the temperature image, different pixels represent areas of different temperatures. Heat dissipation points (hot air pixels) represent the temperature of hot air emitted from the heat source (such as the air outlet of an industrial computer). These pixels have higher pixel values, showing a temperature higher than the ambient temperature. Ambient temperature pixels represent the temperature of the surrounding environment in the temperature image. Their pixel values ​​are usually lower, showing a temperature consistent with the surrounding environment.

[0043] This application will be located in the normal distribution curve of the gradient value ( ) pixels outside the interval can be used as heat dissipation points, or ( ) interval as heat dissipation points, which depends on whether there are more pixels representing heat dissipation points or more pixels representing ambient temperature in the temperature image. That is, when there are more pixels with higher gradient values ​​(or higher pixel values), ( ) interval as heat dissipation points, otherwise, ( ) The pixels outside the interval are regarded as heat dissipation points. The purpose of the above steps is to divide the heat dissipation points and the ambient temperature points.

[0044] S3: Calculate ventilation rating.

[0045] The calculation formula for ventilation evaluation is: ,in, Indicates ventilation rating, Indicates the number of heat dissipation points in the current frame temperature image. Indicates the number of heat dissipation points in the next frame temperature image of the current frame. Represents the set of pixel values ​​of heat dissipation points in the temperature image of the current frame, Represents the set of pixel values ​​of heat dissipation points in the temperature image of the next frame of the current frame, Indicates the average.

[0046] Expressed collection The average value of the pixel value of the heat dissipation point in the middle, Expressed collection The average value of the pixel values ​​of the heat sink.

[0047] Ventilation evaluation is used to measure the heat dissipation efficiency or ventilation effect of the heat dissipation point between two frames of temperature images. By comparing the heat dissipation quantity and the average temperature of the heat dissipation point in the two frames of images, the change in ventilation effect is quantified. The larger the value, the better the ventilation effect becomes over time. The smaller the value, the worse the ventilation effect becomes over time.

[0048] S4: Input the ventilation evaluation and the real-time temperature into a preset prediction model to generate a temperature prediction value, and adjust the heat dissipation level of the industrial computer according to the temperature prediction value, wherein each heat dissipation level corresponds to a prediction model.

[0049] The construction of the prediction model includes: setting a data set, which includes the historical temperature data of the industrial computer and the corresponding ventilation evaluation. The data set is divided into a training set and a validation set, for example, 80% training set and 20% validation set. The ratio of the training set to the validation set can be adjusted according to the actual application scenario. The specific process of model training, such as model training based on the training set and model performance verification based on the validation set, is an existing technology and will not be described in detail here.

[0050] The training set is put into the neural network model for model training until the preset model training stop condition is reached to complete the training and obtain the prediction model.

[0051] In one embodiment, the neural network model may select a TCN (Temporal Convolutional Network) model or an LSTM (Long Short-Term Memory) network model, etc. Both the TCN model and the LSTM network model are existing technologies and will not be described in detail here.

[0052] In one embodiment, the loss function of the model training is a mean square error loss function. The mean square error loss function is a prior art loss function. Other loss functions that can be applied to the application scenarios of this application may also be used, which will not be elaborated here.

[0053] The model training stop condition is that the loss function is less than a preset threshold or the model training times reach a preset number of times. For example, the preset threshold can be 0.01, the preset number of times can be 1000 times, and the preset threshold and the number of predictions can be set by the implementer according to the specific scenario.

[0054] When the real-time temperature average within the preset time is greater than the predicted temperature average, the heat dissipation gear is increased to increase the fan speed. For example, one gear can be increased each time, such as from gear 2 to gear 3. The preset time can be 5s, that is, there are 5 temperature prediction values ​​within the preset time. The preset time can be set according to the actual application scenario;

[0055] When the real-time temperature average within the preset time is equal to the predicted temperature average, the current cooling level is maintained.

[0056] When the real-time temperature average within the preset time is less than the predicted temperature average, the heat dissipation gear is adjusted down to reduce the fan speed. For example, the gear can be lowered one gear at a time, such as from gear 2 to gear 1.

[0057] In one embodiment, when using a neural network model for prediction, the prediction results may be over-dispersed. In order to reduce the number of training times, the loss function is corrected during the training process. The correction method is as follows:

[0058] Since the temperature changes nonlinearly and the temperature change will not be over-discrete in actual application scenarios, a nonlinear change curve is constructed. The construction method is: using the least squares method, the temperature collection values ​​​​at multiple moments before the prediction moment are fitted with a spline curve to obtain a fitted nonlinear curve. For example, the data of the first 20 moments are taken.

[0059] Calculate the distance between the temperature prediction value at the prediction time and the fitted nonlinear curve, and obtain the correction coefficient based on the distance. Specifically, the expression of the correction coefficient is: ,in, Indicates The correction coefficient corresponding to the temperature prediction value is: Indicates The distance between the predicted temperature value and the fitted nonlinear curve, Represents the total number of temperature prediction values.

[0060] The correction coefficient is used as the weight of the loss function to obtain the corrected loss function, and training is performed again. The above steps are repeated until the loss function is less than the preset threshold or the model training times reaches the preset times, and the training of the neural network is completed.

[0061] Taking the first group of temperature prediction values ​​as 5 as an example, Substituting into the loss function, the expression of the loss function is: , represents the loss function, Indicates the correction coefficient corresponding to the first temperature prediction value, Indicates the actual temperature corresponding to the first temperature prediction value, Indicates the first temperature prediction value. Indicates the correction coefficient corresponding to the fifth temperature prediction value, Indicates the actual temperature corresponding to the fifth temperature prediction value, Indicates the fifth temperature prediction value.

[0062] An embodiment of the present application also discloses an industrial computer heat dissipation control system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an industrial computer heat dissipation control method according to the present application is implemented.

[0063] The above system also includes other components such as a communication bus and a communication interface that are well known to technicians in this field. Their settings and functions are known in this field and will not be described in detail here.

[0064] In the present application, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high bandwidth memory HBM (High Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device.

[0065] Although this specification has shown and described a plurality of embodiments of the present application, it is obvious to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will think of many changes, modifications and alternatives without departing from the thought and spirit of the present application. It should be understood that in the process of practicing the present application, various alternatives to the embodiments of the present application described herein may be adopted.

[0066] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A heat dissipation control method for an industrial computer, characterized in that: Includes steps: Obtain temperature images of continuous frames at the air outlet, calculate the gradient values ​​of the pixels in the temperature image, and construct a gradient value set based on the gradient values ​​of each pixel; The gradient value set is input into a preset Gaussian distribution model to generate a Gaussian distribution curve, and the pixel points corresponding to the gradient values ​​outside / inside the preset interval of the Gaussian distribution curve are used as heat dissipation points; Calculate the ventilation rating. The calculation formula for the ventilation rating is: ,in, Indicates ventilation rating, Indicates the number of heat dissipation points in the current frame temperature image. Indicates the number of heat dissipation points in the next frame temperature image of the current frame. Represents the set of pixel values ​​of heat dissipation points in the temperature image of the current frame, Represents the set of pixel values ​​of heat dissipation points in the temperature image of the next frame of the current frame, Indicates the average; The ventilation evaluation and the real-time temperature are input into a preset prediction model to generate a temperature prediction value, and the heat dissipation level of the industrial computer is adjusted according to the temperature prediction value, wherein each heat dissipation level corresponds to a prediction model; The construction of the prediction model includes the following steps: setting a data set, which includes the historical temperature data of the industrial computer and the corresponding ventilation evaluation; dividing the data set into a training set and a validation set, and putting the training set into the neural network model for model training until the preset model training stop condition is reached to complete the training and obtain the prediction model; correcting the loss function, including the following steps: using the least squares method to fit the temperature collection values ​​at multiple moments before the prediction moment with a spline curve to obtain a fitted nonlinear curve; calculating the distance between the temperature prediction value at the prediction moment and the fitted nonlinear curve, and obtaining the correction coefficient according to the distance, and the expression of the correction coefficient is: ,in, Indicates The correction coefficient corresponding to the temperature prediction value is: Indicates The distance between the predicted temperature value and the fitted nonlinear curve, Represents the total number of temperature prediction values; the correction coefficient is used as the weight of the loss function to obtain the corrected loss function.

2. The heat dissipation control method for an industrial computer according to claim 1, characterized in that: The method for calculating the gradient value of a pixel point in a temperature image comprises the following steps: Get the pixel value of any pixel point and the maximum value of the pixel values ​​of the eight neighboring pixels centered on the pixel point, and take the ratio of the maximum value to the pixel value of the pixel point as the gradient value of the pixel point; Traverse all pixels and get the gradient values ​​of all pixels.

3. The heat dissipation control method for an industrial computer according to claim 1, characterized in that: The preset interval is Gaussian distribution ( ) interval, is the mean of the data in the Gaussian distribution, Represents the standard deviation of data from a Gaussian distribution.

4. The heat dissipation control method for an industrial computer according to claim 1, characterized in that: Adjusting the heat dissipation level of the industrial computer according to the temperature prediction value includes the following steps: In response to the real-time temperature average within the preset time being greater than the predicted temperature average, the heat dissipation gear is increased to increase the fan speed; In response to the real-time temperature average within the preset time being less than the predicted temperature average, the heat dissipation gear is reduced to reduce the fan speed; In response to the real-time temperature average within the preset time being equal to the predicted temperature average, the current heat dissipation gear is maintained.

5. An industrial computer heat dissipation control system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the heat dissipation control method for an industrial computer according to any one of claims 1 to 4 is implemented.

Citation Information

Patent Citations

  • Control by temperature change industrial computer case

    CN208207694U

  • Data-based qualitative analysis method of influence factors of circuit breaker faults

    CN107301296A

  • Wind speed vector prediction method based on Gaussian process and combined model

    CN115907099A

  • Big data-based sales information prediction method and device

    CN116596582A

  • Sea surface temperature image completion method and network based on de-noising diffusion generation model

    CN117036880A