Industrial personal computer control method and system based on artificial intelligence

Abstract

The invention provides an industrial personal computer control method and system based on artificial intelligence, and the method comprises the steps: obtaining a key heating assembly parameter of an industrial personal computer, and constructing a heat impact comprehensive index of the industrial personal computer; establishing a heat dissipation demand prediction model based on the heat influence comprehensive index of the industrial personal computer; formulating a heat dissipation control strategy according to a comparison result of the predicted heat influence comprehensive index and a preset heat influence comprehensive index safety threshold value; and according to the heat dissipation control strategy, predicting the heat influence comprehensive index after heat dissipation, and if the actual heat influence comprehensive index is greater than the predicted heat influence comprehensive index after heat dissipation, generating prompt information for cleaning dust of the heat dissipation fan of the industrial personal computer. By means of the method and the corresponding system, abrasion of the fan motor of the industrial personal computer can be reduced, the service life of the fan of the industrial personal computer is prolonged, and a user is helped to find and solve the heat dissipation problem caused by dust deposition of the fan in time.

Description

Technical Field

[0001] The invention proposes an industrial computer control method and system based on artificial intelligence, and relates to the field of heat dissipation control. Background Art

[0002] Industrial computers are computer equipment specially designed for industrial environments. Industrial scenes are often accompanied by high temperatures, dust, and poor power supply conditions, and industrial computers need to operate continuously for a long time in such environments. In this case, whether it has an excellent heat dissipation design has become a key factor in ensuring the stable operation of the industrial computer.

[0003] The traditional IPC cooling control method adjusts the fan speed based on predictions in a one-size-fits-all manner, and cannot adjust the fan speed based on the IPC's own operating conditions. This one-size-fits-all adjustment instead of adjusting it based on the IPC's own operating conditions sometimes increases the wear of the fan motor and shortens the fan's service life. Because the current and mechanical stress on the motor change greatly during the fan's startup, acceleration, and deceleration, it is prone to failure in the long term. Furthermore, sometimes due to the IPC's own hardware problems, such as dust accumulation on the cooling fan, no matter how the fan speed is adjusted, it is useless. It is impossible to judge whether there is an abnormality in the IPC's own cooling system based on the IPC's own operating status. Summary of the invention

[0004] The present invention provides an industrial computer control method and system based on artificial intelligence, which predicts the heat dissipation demand and adjusts the predicted heat dissipation cycle. Furthermore, based on the heat dissipation cycle, a cleaning trigger coefficient calculated by a cleaning trigger coefficient model is used to judge whether the industrial computer's own hardware has a heat dissipation failure to solve the above-mentioned problems.

[0005] The present invention proposes an industrial computer control method based on artificial intelligence, the method comprising: Obtain the parameters of key heat-generating components of industrial computers and construct comprehensive thermal impact indicators of industrial computers; Establish a heat dissipation demand prediction model based on the comprehensive thermal impact index of industrial computers; Formulate a heat dissipation control strategy based on the comparison results between the predicted comprehensive heat impact index and the preset safety threshold of the comprehensive heat impact index; According to the heat impact comprehensive index after heat dissipation predicted by the heat dissipation control strategy, if the actual heat impact comprehensive index is greater than the predicted heat impact comprehensive index after heat dissipation, a prompt message for cleaning dust from the heat dissipation fan of the industrial computer is generated.

[0006] Furthermore, the parameters of the key heat-generating components of the industrial computer are obtained to construct a comprehensive index of the thermal impact of the industrial computer, including: Obtaining parameters of key heat generating components of the industrial computer, including: CPU temperature, GPU temperature, memory chip temperature, heat sink temperature, temperature difference between both ends of the heat pipe, fan speed, GPU usage, GPU load, CPU usage and CPU load; Record the timestamp of each data collection for subsequent time series analysis; The thermal impact comprehensive index of the industrial computer is constructed by weighted summation of the collected key heat-generating component parameters, and the weight coefficient of the key heat-generating component parameters is determined according to the performance of the industrial computer.

[0007] Furthermore, the weight coefficients of the key heating component parameters are determined according to the performance of the industrial computer, including: The process of determining the weight coefficient of the comprehensive heat impact index is divided into a target layer, a criterion layer and an indicator layer. The target layer is to determine the weight coefficient of the comprehensive heat impact index. The criterion layer includes CPU performance, GPU performance, memory performance, heat sink cooling capacity, heat pipe cooling capacity, internal environment impact and external environment impact. The indicator layer corresponds to the temperature parameter. For industrial control hosts with different performance, the relative importance of each criterion is analyzed, the host performance parameters are quantitatively analyzed, and a judgment matrix is ​​constructed; The maximum eigenvalue of the judgment matrix and its corresponding eigenvector are calculated using the eigenroot method. After normalization, the eigenvector is the weight vector of each criterion relative to the target layer. The judgment matrix is ​​checked for consistency, the consistency index is calculated, and the corresponding average random consistency index is found. Then the consistency ratio is calculated. When the consistency ratio is less than 0.1, the judgment matrix is ​​considered to have satisfactory consistency, otherwise the judgment matrix needs to be readjusted.

[0008] Furthermore, a heat dissipation demand prediction model is established based on the comprehensive thermal impact index of the industrial computer, including: Acquire a historical data set, wherein the historical data set includes historical key heat-generating component parameters of the industrial computer and corresponding comprehensive heat impact indicators; The historical data set is divided into a training set and a validation set, and the training set is put into the neural network model for training until the set model training end conditions are met, and finally a heat dissipation demand prediction model is obtained; The prediction period is calculated, and when the time reaches the prediction period, the temperature of the industrial computer is predicted by the established heat dissipation demand prediction model.

[0009] Furthermore, the prediction period is calculated, including: The prediction period is calculated by using a prediction period model. Specifically, the prediction period model is: ; in, represents the adjusted forecast period, represents the basic forecast period, Indicates the difference between the current and previous CPU usage. Indicates the difference between the GPU usage at the current moment and the previous moment. Indicates the average CPU usage during the preset time period. represents the average value of GPU usage in the past preset time period, β represents the first weight coefficient, and σ represents the second weight coefficient.

[0010] Furthermore, a heat dissipation control strategy is formulated based on the comparison result between the predicted comprehensive heat impact index and the preset safety threshold of the comprehensive heat impact index, including: Comparing the predicted heat dissipation impact comprehensive index with the preset heat impact comprehensive index; If the predicted comprehensive heat dissipation impact index is greater than the preset comprehensive heat impact index, the speed of the heat dissipation fan is adjusted through a control algorithm.

[0011] Further, according to the heat impact comprehensive index after heat dissipation predicted by the heat dissipation control strategy, if the actual heat impact comprehensive index is greater than the predicted heat impact comprehensive index after heat dissipation, a prompt message for cleaning dust from the heat dissipation fan of the industrial computer is generated, including: Obtain the predicted comprehensive index value of thermal impact after heat dissipation; Use the temperature sensor installed on the industrial computer to obtain the actual thermal impact comprehensive index value of the current industrial computer after heat dissipation; According to the predicted comprehensive index value of heat impact after heat dissipation and the actual comprehensive index value of heat impact after heat dissipation, the cleaning trigger coefficient K is calculated with the help of the cleaning trigger coefficient model. When the cleaning trigger coefficient K is greater than the first preset threshold and When it is greater than zero, a gradient warning signal is generated; If the gradient warning signal continues for 3 prediction cycles, a prompt message for fan cleaning is output to the display of the industrial computer.

[0012] The present invention proposes an industrial computer control system based on artificial intelligence, the system comprising: Construct a comprehensive heat impact index module to obtain the parameters of key heat-generating components of industrial computers and construct comprehensive heat impact indexes for industrial computers; Establish a heat dissipation demand prediction model module, which is used to establish a heat dissipation demand prediction model based on the comprehensive thermal impact index of the industrial computer; A heat dissipation control module, used to formulate a heat dissipation control strategy according to a comparison result between the predicted heat impact comprehensive index and a preset heat impact comprehensive index safety threshold; The prompt module is used to generate a prompt message for cleaning dust from the industrial computer cooling fan if the actual comprehensive heat impact index is greater than the predicted comprehensive heat impact index after cooling according to the heat dissipation control strategy.

[0013] Furthermore, an industrial computer control system based on artificial intelligence includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the industrial computer control method based on artificial intelligence when executing the program.

[0014] A computer program product proposed in the present invention includes a computer program and / or instructions, and is characterized in that when the computer program and / or instructions are executed by a processor, the steps of the industrial computer control method based on artificial intelligence are implemented.

[0015] The beneficial effects of the present invention are as follows: the comprehensive heat impact index comprehensively considers various factors such as the internal heating components, heat dissipation components and internal and external environmental temperatures of the industrial computer, and can accurately evaluate the heat dissipation condition of the industrial computer, avoid the one-sidedness of judging the heat dissipation condition based on a single temperature parameter, and provide an accurate data basis for subsequent heat dissipation control; the heat dissipation demand prediction model based on a large amount of historical data training can accurately predict the comprehensive heat impact index according to the current load condition and real-time parameters such as the fan speed, predict the heat dissipation demand in advance, reduce the lag of heat dissipation control, and improve the dynamic response speed of the temperature control system. According to the comparison between the prediction result and the safety threshold, the heat dissipation control strategy of adjusting the fan speed by using the PID control algorithm is realized, and intelligent adjustment of heat dissipation is realized; by comparing the actual and predicted comprehensive heat impact indexes, when the actual index is greater than the predicted index, a prompt message for cleaning the fan dust is generated in time, and an early warning of heat dissipation failure is realized, which helps users to promptly discover and solve the heat dissipation problem caused by fan dust accumulation, and avoid the more serious failure of the industrial computer performance degradation caused by poor heat dissipation and the industrial computer component damage caused by excessive temperature. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a schematic diagram of an industrial computer control method based on artificial intelligence described in the present invention. DETAILED DESCRIPTION

[0017] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0018] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. The embodiments described are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0020] One embodiment of the present invention provides an industrial computer control method based on artificial intelligence, the method comprising: Obtain the parameters of key heat-generating components of industrial computers and construct comprehensive thermal impact indicators of industrial computers; Establish a heat dissipation demand prediction model based on the comprehensive thermal impact index of industrial computers; Formulate a heat dissipation control strategy based on the comparison results between the predicted comprehensive heat impact index and the preset safety threshold of the comprehensive heat impact index; According to the heat impact comprehensive index after heat dissipation predicted by the heat dissipation control strategy, if the actual heat impact comprehensive index is greater than the predicted heat impact comprehensive index after heat dissipation, a prompt message for cleaning dust from the heat dissipation fan of the industrial computer is generated.

[0021] The working principle and effect of the above technical solution are as follows: the comprehensive heat impact index comprehensively considers various factors such as the internal heat-generating components, heat dissipation components and internal and external environmental temperatures of the industrial computer, and can accurately evaluate the heat dissipation status of the industrial computer, avoiding the one-sidedness of judging the heat dissipation status based on a single temperature parameter, and providing an accurate data basis for subsequent heat dissipation control; the heat dissipation demand prediction model based on a large amount of historical data training can accurately predict the comprehensive heat impact index according to the current load status and real-time parameters such as fan speed, predict the heat dissipation demand in advance, reduce the lag of heat dissipation control, and improve the dynamic response speed of the temperature control system. According to the comparison between the prediction result and the safety threshold, the PID control algorithm is used to adjust the heat dissipation control strategy of the fan speed, and the intelligent adjustment of heat dissipation is realized; by comparing the actual and predicted comprehensive heat impact indexes, when the actual index is greater than the predicted index, a prompt message for cleaning the fan dust is generated in time, and an early warning of heat dissipation failure is realized, which helps users to promptly discover and solve the heat dissipation problem caused by fan dust accumulation, and avoid the more serious failure of the industrial computer performance degradation caused by poor heat dissipation and the industrial computer component damage caused by excessive temperature.

[0022] In one embodiment of the present invention, parameters of key heat-generating components of an industrial computer are obtained to construct a comprehensive index of thermal impact of the industrial computer, including: Obtaining parameters of key heat generating components of the industrial computer, including: CPU temperature, GPU temperature, memory chip temperature, heat sink temperature, temperature difference between both ends of the heat pipe, fan speed, GPU usage, GPU load, CPU usage and CPU load; Get the CPU temperature through the temperature sensor integrated inside the CPU; Get the GPU temperature through the temperature sensor attached to the GPU surface; Monitor the memory chip temperature through the temperature sensor located at the memory chip on the memory module; Monitor the heat sink temperature by a temperature sensor installed at the top of the heat sink fins; The temperature difference between the two ends of the heat pipe is monitored by temperature sensors installed at both ends of the heat pipe; The internal and external ambient temperatures of the industrial computer are collected through temperature sensors; Use hardware monitoring tools to obtain fan speed, GPU usage, GPU load, CPU usage, and CPU load data; Record the timestamp of each data collection for subsequent time series analysis; The thermal impact comprehensive index of the industrial computer is constructed by weighted summation of the collected key heat-generating component parameters, and the weight coefficient of the key heat-generating component parameters is determined according to the performance of the industrial computer.

[0023] The temperature sensors are used to collect the temperatures of key heat generating components such as CPU and GPU memory, recorded as At the same time, sensors are set at the joint between the heat sink and the heating component and on the top of the fins to obtain the heat sink temperature. Install sensors at both ends of the heat pipe to record the temperature at both ends of the heat pipe. , collect the internal ambient temperature of the industrial computer through the ambient temperature sensor And the external ambient temperature , use the hardware monitoring tool to obtain the fan speed And the CPU usage of the IPC running the program , GPU usage Load data.

[0024] ; Among them, H represents the comprehensive index of thermal impact, and They represent weight coefficients respectively.

[0025] The working principle and effect of the above technical solution are: the key heating component parameters such as CPU temperature, GPU temperature, memory particle temperature, heat sink temperature, temperature difference at both ends of the heat pipe, fan speed, GPU utilization rate, GPU load, CPU utilization rate, CPU load and internal and external ambient temperature collected are used to construct a comprehensive thermal impact index of the industrial computer by weighted summation. The weight coefficient of the key heating component is not fixed, but is determined according to the performance of the industrial computer. For industrial computers mainly based on CPU calculations, the weight coefficient of CPU-related parameters (such as CPU temperature, CPU utilization rate, CPU load) will be relatively large; for industrial computers focusing on graphics processing, the weight coefficient of GPU-related parameters is higher. By determining the weight coefficient of the key heating component according to the performance of the industrial computer, it can more accurately reflect the actual thermal impact of industrial computers with different performance under various working conditions, and provide accurate data support for subsequent heat dissipation analysis and control.

[0026] In one embodiment of the present invention, the weight coefficient of the key heating component parameter is determined according to the performance of the industrial computer, including: The process of determining the weight coefficient of the comprehensive heat impact index is divided into a target layer, a criterion layer and an indicator layer. The target layer is to determine the weight coefficient of the comprehensive heat impact index. The criterion layer includes CPU performance, GPU performance, memory performance, heat sink cooling capacity, heat pipe cooling capacity, internal environment impact and external environment impact. The indicator layer corresponds to the temperature parameter. For industrial control hosts with different performance, the relative importance of each criterion is analyzed, the host performance parameters are quantitatively analyzed, and a judgment matrix is ​​constructed; The maximum eigenvalue of the judgment matrix and its corresponding eigenvector are calculated using the eigenroot method. After normalization, the eigenvector is the weight vector of each criterion relative to the target layer. The judgment matrix is ​​checked for consistency, the consistency index is calculated, and the corresponding average random consistency index is found. Then the consistency ratio is calculated. When the consistency ratio is less than 0.1, the judgment matrix is ​​considered to have satisfactory consistency, otherwise the judgment matrix needs to be readjusted.

[0027] The working principle and effect of the above technical solution are as follows: through the structured process of hierarchical analysis method, various factors affecting the comprehensive index of thermal impact are taken into account, and the analysis is gradually refined from different levels, so that the determination of weight coefficients is more scientific and reasonable; the relative importance of each criterion is analyzed for industrial control hosts with different performances, so that the weight determination method has strong versatility and pertinence. Whether it is an industrial control host mainly based on CPU calculations or a host focusing on GPU graphics processing, the weight vector that meets its performance characteristics can be accurately determined by adjusting the judgment matrix construction process, thereby providing a practical basis for subsequent heat dissipation analysis and control; it avoids the situation where the weight coefficient is unreasonable due to inconsistent judgment matrices, making the determination process of the weight coefficient of the comprehensive index of thermal impact more rigorous, laying a solid foundation for subsequent thermal analysis and heat dissipation strategy formulation based on the weight coefficient, and improving the stability and reliability of the entire industrial control computer heat dissipation control method.

[0028] In one embodiment of the present invention, a heat dissipation demand prediction model is established based on the comprehensive heat impact index of the industrial computer, including: Acquire a historical data set, wherein the historical data set includes historical key heat-generating component parameters of the industrial computer and corresponding comprehensive heat impact indicators; The historical data set is divided into a training set and a validation set, and the training set is put into the neural network model for training until the set model training end conditions are met, and finally a heat dissipation demand prediction model is obtained; The prediction period is calculated, and when the time reaches the prediction period, the temperature of the industrial computer is predicted by the established heat dissipation demand prediction model.

[0029] In one embodiment of the present invention, the calculation of the prediction period includes: In order to obtain the absolute performance change index of the industrial computer, the difference between the CPU usage at the current moment and the previous moment is obtained, and then the difference between the GPU usage at the current moment and the previous moment is obtained. The sum of the squares of the two differences is calculated, and then the square root of the sum of the squares is calculated. , which represents the modulus of the "change vector" in two-dimensional space, represents the comprehensive intensity of the change in CPU and GPU usage. Regardless of the direction of change in CPU and GPU usage (positive or negative), only the amplitude is concerned, and the square root is multiplied by the weight coefficient β to obtain the comprehensive intensity of the change in CPU and GPU usage. By adjusting β, the weight of the absolute fluctuation on the prediction period can be controlled. Therefore, the evaluation coefficient of the short-term drastic fluctuation intensity of industrial computer resource usage, that is, the instantaneous fluctuation intensity, can be expressed as: ; Get the average value of CPU usage in the past preset time period, divide the difference between the current and previous CPU usage by it to get the quotient, then add it to the difference between the current and previous GPU usage and the average value of GPU usage in the past preset time period, take the absolute value as a whole to get the proportional change relative to the historical mean, and then multiply it by the weight coefficient σ to capture the long-term trend deviation. The larger σ is, the more sensitive the system is to relative deviation. For example, in scenarios where the resource usage baseline is unstable (such as cloud computing environments), σ can be increased to emphasize trend changes; in scenarios where the baseline is stable (such as embedded systems), σ can be reduced to rely mainly on absolute fluctuations. Therefore, the deviation of the resource fluctuation of the industrial computer from the long-term average level, that is, the evaluation coefficient of the long-term trend deviation, can be expressed as: ; In order to achieve more comprehensive and robust adaptive control, the evaluation coefficient of instantaneous fluctuation intensity is added to the evaluation coefficient of long-term trend deviation to obtain the overall evaluation coefficient, namely: ; In order to avoid over-adjustment of the prediction period, we then obtain the difference between the GPU usage at the current moment and the previous moment, calculate the sum of the squares of the two differences, and then calculate the square root of the sum of the squares, which is , reduce the impact of the β term on the prediction period, that is, the impact of excessive increase, obtain the average value of the CPU usage in the past preset time period, divide the difference between the current moment and the previous moment CPU usage by it to obtain the quotient, and then add it to the difference between the current moment and the previous moment GPU usage and the average value of the GPU usage in the past preset time period. Take the absolute value as a whole to reduce the impact of the σ term on the prediction period, that is, the impact of excessive increase, and combine the parts that suppress the β term and the σ term to obtain the overall suppression coefficient, that is: ; In order to avoid the denominator being 0, and when there is no fluctuation , the denominator degenerates to 1, which is in line with the basic logic and represents the default stable state of the system (no adjustment is required when there is no fluctuation). The overall suppression coefficient is added with 1, and the overall evaluation coefficient is combined with the overall suppression coefficient to obtain the initial period adjustment coefficient, namely: ; In order to convert the intensity of fluctuations into the compression coefficient of the forecast period, the stronger the fluctuation, the shorter the period, and obtain the overall period adjustment coefficient, that is: ; Then the final expression of the forecast period model can be expressed as: ; in, represents the adjusted forecast period, represents the basic forecast period, Indicates the difference between the current and previous CPU usage. Indicates the difference between the GPU usage at the current moment and the previous moment. Indicates the average CPU usage during the preset time period. represents the average value of GPU usage in the past preset time period, β represents the first weight coefficient, and σ represents the second weight coefficient.

[0030] The working principle and effect of the above technical solution are: the prediction cycle can be adjusted according to the real-time status of the industrial computer. and Incorporating it into the formula and calculating the difference in change by taking the square root of the square root can fully reflect the fluctuation degree of CPU and GPU usage. Any significant change in either side can be captured, providing a basis for adjusting the prediction period. The greater the fluctuation degree of CPU and GPU usage, the shorter the prediction period. Considering the change rate and , combined with the average usage rate, it can be determined whether the current change deviates from the normal state. If the CPU usage change value is small in a short period of time, but the change rate is large compared with the average usage rate, it also indicates that the load is unstable. At this time, it is also necessary to adjust the prediction period. The greater the CPU and GPU usage deviates from the average usage rate, the shorter the prediction period. The weight coefficients β and σ can flexibly adjust the influence of the change difference and the change rate on the prediction period according to the actual situation, and enhance the adaptability of the formula; and by normalizing the denominator and constructing the overall formula in the inverse form, it ensures that the adjustment factor is reasonable. When the load changes greatly, the prediction period is shortened rapidly, and the heat dissipation demand is followed up in time; when the load is stable, the prediction period is extended to reduce unnecessary calculations and improve the overall operation efficiency of the system; avoid frequently adjusting the fan speed according to the prediction, and adjust the fan speed only when necessary, reduce the wear of the fan motor, and extend the service life of the fan.

[0031] In one embodiment of the present invention, a heat dissipation control strategy is formulated according to a comparison result between a predicted comprehensive heat impact index and a preset safety threshold of the comprehensive heat impact index, including: Comparing the predicted heat dissipation impact comprehensive index with the preset heat impact comprehensive index; If the predicted comprehensive heat dissipation impact index is greater than the preset comprehensive heat impact index, the speed of the heat dissipation fan is adjusted through a control algorithm.

[0032] The calculation formula for the fan speed adjustment ΔR is: ; in, and are proportional, integral and differential coefficients respectively, It represents the comprehensive index of predicted thermal impact, It represents the preset safety threshold of the comprehensive thermal impact index, ΔR represents the fan speed adjustment amount, and t represents the time variable, which represents the time from the beginning of the prediction period to the current moment.

[0033] The working principle and effect of the above technical solution are: timely responding to heat dissipation needs, through real-time comparison and prediction of comprehensive thermal impact indicators and safety thresholds, and immediately starting the PID control algorithm when the threshold is exceeded, it can quickly respond to the heat dissipation needs of the industrial computer, improve the heat dissipation response speed, and take timely measures in the early stage of thermal problems, reduce the lag of heat dissipation needs, and avoid performance degradation or hardware damage of the industrial computer due to untimely heat dissipation.

[0034] In one embodiment of the present invention, according to the comprehensive heat impact index after heat dissipation predicted by the heat dissipation control strategy, if the actual comprehensive heat impact index is greater than the predicted comprehensive heat impact index after heat dissipation, a prompt message for cleaning dust from the heat dissipation fan of the industrial computer is generated, including: Obtain the predicted comprehensive index value of thermal impact after heat dissipation; Use the temperature sensor installed on the industrial computer to obtain the actual thermal impact comprehensive index value of the current industrial computer after heat dissipation; According to the predicted comprehensive index value of heat impact after heat dissipation and the actual comprehensive index value of heat impact after heat dissipation, the cleaning trigger coefficient K is calculated with the help of the cleaning trigger coefficient model. When the cleaning trigger coefficient K is greater than the first preset threshold and When it is greater than zero, a gradient warning signal is generated; If the gradient warning signal continues for 3 prediction cycles, a prompt message for fan cleaning is output to the display of the industrial computer.

[0035] The deviation between the actual heat impact index and the predicted value is divided by the predicted heat impact comprehensive index after heat dissipation to normalize the instantaneous heat dissipation deviation, reflecting the relative proportion of the current heat dissipation effect deviating from the predicted value, and then multiplied by the weight α to adjust the influence weight of the current deviation on the trigger coefficient K to obtain the instantaneous heat dissipation deviation factor. Its specific expression can be expressed as: ; The integral in quantifies the fan's continuous adjustment efforts during the cycle. If the fan runs at high speed for a long time (Δω is high), it indicates that the cooling system is under great pressure and needs to be triggered to reduce the load. The exponential function makes the recent adjustment have a greater impact than the early adjustment ( The larger it is, the faster the old data decays), which is consistent with the engineering intuition that “the more recent the behavior, the more relevant it is”; This is to ensure that the speed increments of fans of different models are comparable (for example, fans rated at 5000RPM and 10000RPM have the same physical meaning of the same proportional increments). Therefore, the expression for the speed cumulative load factor can be expressed as: ; A high frequency of ΔH > 0 indicates that the system is in a critical heat dissipation state for a long time, which may imply problems such as insufficient heat dissipation design or dust accumulation, and needs to be cleaned in advance. The "last J hours" focuses on short-term history to avoid interference from old data (such as anomalies from a week ago are irrelevant to the current data). The ratio (rather than the absolute number) is used to avoid deviations caused by different sampling frequencies (such as sampling every minute vs. sampling every hour). Therefore, the expression of the historical anomaly frequency factor can be expressed as: ; Therefore, the expression of the cleaning trigger coefficient model can be expressed as: ; in, That is, the difference between the actual thermal impact comprehensive index value and the predicted thermal index value. In order to predict the comprehensive index of thermal impact after heat dissipation, is the maximum rated speed of the fan, for The fan speed increment at any time. For the last J hours The number of times, is the total number of sampling times, and is the weighting coefficient, is the speed influence attenuation factor, Indicates the starting time of the forecast period, which is different from the current time The time zone of integration is determined together .

[0036] The working principle and effect of the above technical solution are: Instant thermal deviation term Relative deviation using rather than absolute values, eliminating the dimensional differences under different working conditions; when When the heat dissipation is expected to be good, the same The contribution value is greater, strengthening low-load anomaly detection; it can quickly respond to sudden cooling failures (such as fan stoppage) to avoid misjudgment in high-load scenarios (because it allows larger absolute temperature fluctuations); long-term load accumulation items middle Quantify the cumulative speed increment, reflecting the continuous overload operation; through exponential decay Achieve nonlinear saturation characteristics to avoid infinite accumulation; can capture gradual degradation (such as fan efficiency degradation caused by slow accumulation of dust); Adjust the historical impact weight ( When increasing, it focuses on recent data); Measure the statistical probability of abnormal events, identify intermittent faults (such as occasional poor heat dissipation caused by dust), and suppress single false triggers (multiple abnormalities are required to significantly increase the K value).

[0037] In one embodiment of the present invention, the industrial computer control system based on artificial intelligence includes: Construct a comprehensive heat impact index module to obtain the parameters of key heat-generating components of industrial computers and construct comprehensive heat impact indexes for industrial computers; Establish a heat dissipation demand prediction model module, which is used to establish a heat dissipation demand prediction model based on the comprehensive thermal impact index of the industrial computer; A heat dissipation control module, used to formulate a heat dissipation control strategy according to a comparison result between the predicted heat impact comprehensive index and a preset heat impact comprehensive index safety threshold; The prompt module is used to generate a prompt message for cleaning dust from the industrial computer cooling fan if the actual comprehensive heat impact index is greater than the predicted comprehensive heat impact index after cooling according to the heat dissipation control strategy.

[0038] One embodiment of the present invention is an industrial computer control system based on artificial intelligence, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that when the processor executes the program, the steps of the industrial computer control method based on artificial intelligence are implemented.

[0039] One embodiment of the present invention is a computer program product, including a computer program and / or instructions, characterized in that when the computer program and / or instructions are executed by a processor, the steps of the industrial computer control method based on artificial intelligence are implemented.

[0040] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An industrial computer control method based on artificial intelligence, characterized in that: The method comprises: Obtain the parameters of key heat-generating components of industrial computers and construct comprehensive thermal impact indicators of industrial computers; Establish a heat dissipation demand prediction model based on the comprehensive thermal impact index of industrial computers; Formulate a heat dissipation control strategy based on the comparison results between the predicted comprehensive heat impact index and the preset safety threshold of the comprehensive heat impact index; According to the heat impact comprehensive index after heat dissipation predicted by the heat dissipation control strategy, if the actual heat impact comprehensive index is greater than the predicted heat impact comprehensive index after heat dissipation, a prompt message for cleaning dust from the heat dissipation fan of the industrial computer is generated.

2. According to the artificial intelligence-based industrial computer control method of claim 1, it is characterized in that: Obtain the parameters of the key heat-generating components of the industrial computer and construct the comprehensive thermal impact index of the industrial computer, including: Obtaining parameters of key heat generating components of the industrial computer, including: CPU temperature, GPU temperature, memory chip temperature, heat sink temperature, temperature difference between both ends of the heat pipe, fan speed, GPU usage, GPU load, CPU usage and CPU load; Record the timestamp of each data collection for subsequent time series analysis; The thermal impact comprehensive index of the industrial computer is constructed by weighted summation of the collected key heat-generating component parameters, and the weight coefficient of the key heat-generating component parameters is determined according to the performance of the industrial computer.

3. According to the artificial intelligence-based industrial computer control method of claim 2, the weight coefficient of the key heating component parameter is determined according to the performance of the industrial computer, including: The process of determining the weight coefficient of the comprehensive heat impact index is divided into a target layer, a criterion layer and an indicator layer. The target layer is to determine the weight coefficient of the comprehensive heat impact index. The criterion layer includes CPU performance, GPU performance, memory performance, heat sink cooling capacity, heat pipe cooling capacity, internal environment impact and external environment impact. The indicator layer corresponds to the temperature parameter. For industrial control hosts with different performance, the relative importance of each criterion is analyzed, the host performance parameters are quantitatively analyzed, and a judgment matrix is ​​constructed; The maximum eigenvalue of the judgment matrix and its corresponding eigenvector are calculated using the eigenroot method. After normalization, the eigenvector is the weight vector of each criterion relative to the target layer. The judgment matrix is ​​checked for consistency, the consistency index is calculated, and the corresponding average random consistency index is found. Then the consistency ratio is calculated. When the consistency ratio is less than 0.1, the judgment matrix is ​​considered to have satisfactory consistency, otherwise the judgment matrix needs to be readjusted.

4. According to the artificial intelligence-based industrial computer control method of claim 1, it is characterized in that: A heat dissipation demand prediction model is established based on the comprehensive thermal impact indicators of industrial computers, including: Acquire a historical data set, wherein the historical data set includes historical key heat-generating component parameters of the industrial computer and corresponding comprehensive heat impact indicators; The historical data set is divided into a training set and a validation set, and the training set is put into the neural network model for training until the set model training end conditions are met, and finally a heat dissipation demand prediction model is obtained; The prediction period is calculated, and when the time reaches the prediction period, the temperature of the industrial computer is predicted by the established heat dissipation demand prediction model.

5. The method for controlling an industrial computer based on artificial intelligence according to claim 4, characterized in that: The forecast period is calculated by a forecast period model.

6. The method for controlling an industrial computer based on artificial intelligence according to claim 1, characterized in that: According to the comparison results of the predicted comprehensive heat impact index and the preset safety threshold of the comprehensive heat impact index, a heat dissipation control strategy is formulated, including: Comparing the predicted heat dissipation impact comprehensive index with the preset heat impact comprehensive index; If the predicted comprehensive heat dissipation impact index is greater than the preset comprehensive heat impact index, the speed of the heat dissipation fan is adjusted through a control algorithm.

7. The method for controlling an industrial computer based on artificial intelligence according to claim 1, characterized in that: According to the heat impact comprehensive index after heat dissipation predicted by the heat dissipation control strategy, if the actual heat impact comprehensive index is greater than the predicted heat impact comprehensive index after heat dissipation, a prompt message for cleaning dust from the heat dissipation fan of the industrial computer is generated, including: Obtain the predicted comprehensive index value of thermal impact after heat dissipation; Use the temperature sensor installed on the industrial computer to obtain the actual thermal impact comprehensive index value of the current industrial computer after heat dissipation; According to the predicted heat impact comprehensive index value after heat dissipation and the actual heat impact comprehensive index value after heat dissipation, the cleaning trigger coefficient K is calculated with the help of the cleaning trigger coefficient model. When the cleaning trigger coefficient K is greater than the first preset threshold and When it is greater than zero, a gradient warning signal is generated; If the gradient warning signal continues for 3 prediction cycles, a prompt message for fan cleaning is output to the display of the industrial computer.

8. An industrial computer control system based on artificial intelligence, characterized in that: The system comprises: Construct a comprehensive heat impact index module to obtain the parameters of key heat-generating components of industrial computers and construct comprehensive heat impact indexes for industrial computers; Establish a heat dissipation demand prediction model module, which is used to establish a heat dissipation demand prediction model based on the comprehensive thermal impact index of the industrial computer; A heat dissipation control module, used to formulate a heat dissipation control strategy according to a comparison result between the predicted heat impact comprehensive index and a preset heat impact comprehensive index safety threshold; The prompt module is used to generate a prompt message for cleaning dust from the industrial computer cooling fan if the actual comprehensive heat impact index is greater than the predicted comprehensive heat impact index after cooling according to the heat dissipation control strategy.

9. An industrial computer control system based on artificial intelligence, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product comprising a computer program and / or instructions, characterized in that: When the computer program and / or the instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images