Intelligent adjusting method and device of cooling fan intelligent control system capable of resisting sand and dust environment

Abstract

The invention relates to the technical field of intelligent control, and discloses an intelligent adjusting method and device for a cooling fan intelligent control system resistant to a sand and dust environment, and the method comprises the steps: carrying out the multi-working-condition data collection of a cooling test platform of a cooling fan, and obtaining a sand and dust environment sample library; performing quantitative modeling analysis on dust accumulation dynamic characteristics and an air duct blockage formation mechanism based on the sand and dust environment sample library to obtain a thermal equivalent model; performing feature extraction and integration based on the thermal equivalent model to obtain a multi-dimensional dust environment heat dissipation feature vector; the multi-dimensional sand and dust environment heat dissipation feature vectors are input into a gradient lifting learning model with a multi-attraction mechanism to be processed, and fan cooperative control parameters adapting to the sand and dust environment are obtained.According to the method, the working state of a heat dissipation system in the sand and dust environment is comprehensively represented, the problem that description of traditional features in the sand and dust environment is insufficient is solved, and the method is suitable for being applied to the sand and dust environment. Therefore, the adaptive capacity and the reliability of the cooling fan in a complex sand and dust environment are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and in particular to an intelligent adjustment method and device for an intelligent control system of a heat dissipation fan in an anti-dust environment. Background Art

[0002] In industrial environments with severe dust, cooling fans are key components for equipment cooling, but dust accumulation and particle blockage can significantly reduce heat dissipation performance, increase energy consumption, and shorten equipment life. Traditional heat dissipation control methods mainly use iterative search or simple temperature feedback mechanisms to determine fan control parameters. These methods can meet the needs under normal conditions, but in dusty environments, there are problems such as low control success rate, long response time, and inability to cope with dust interference, making it difficult to meet the actual needs of industrial applications.

[0003] The existing heat dissipation control methods usually lack an in-depth understanding of the dynamic characteristics of dust accumulation and the formation mechanism of air duct blockage, and cannot accurately predict the attenuation law of fan performance in a dusty environment, resulting in low control accuracy, high energy consumption, and short fan life. In particular, in a multi-fan system, each fan usually works independently of each other and lacks coordination. The contradiction between dust protection and heat dissipation efficiency is more prominent, the overall performance of the system is difficult to guarantee, and it is difficult to respond promptly and effectively when the dust concentration changes suddenly. Summary of the invention

[0004] The present invention provides an intelligent adjustment method and device for an intelligent control system of a cooling fan resistant to dusty environments. The present invention comprehensively characterizes the working state of the cooling system in a dusty environment, overcomes the problem that traditional characteristics are insufficient in describing the dusty environment, and thereby improves the adaptability and reliability of the cooling fan in a complex dusty environment.

[0005] In a first aspect, the present invention provides an intelligent adjustment method for an intelligent control system of a heat dissipation fan in an anti-sand and dust environment, and the intelligent adjustment method for an intelligent control system of a heat dissipation fan in an anti-sand and dust environment comprises: Collect data from multiple working conditions on the cooling fan's heat dissipation test platform to obtain a dust environment sample library; Based on the dust environment sample library, the dust accumulation dynamic characteristics and the air duct blockage formation mechanism are quantitatively modeled and analyzed to obtain a thermal equivalent model; Based on the thermal equivalent model, feature extraction and integration are performed to obtain a multi-dimensional dust environment heat dissipation feature vector; The multi-dimensional dust environment heat dissipation feature vector is input into a gradient boosting learning model with a multi-attraction mechanism for processing, so as to obtain fan cooperative control parameters adapted to the dust environment.

[0006] In a second aspect, the present invention provides an intelligent control system intelligent adjustment device for a heat dissipation fan in an anti-dust environment, and the intelligent control system intelligent adjustment device for a heat dissipation fan in an anti-dust environment comprises: The acquisition module is used to collect data of multiple working conditions on the heat dissipation test platform of the heat dissipation fan to obtain a dust environment sample library; A modeling module, used to perform quantitative modeling and analysis on the dust accumulation dynamics characteristics and the air duct blockage formation mechanism based on the dust environment sample library to obtain a thermal equivalent model; A feature extraction module, used for extracting and integrating features based on the thermal equivalent model to obtain a multi-dimensional dust environment heat dissipation feature vector; The processing module is used to input the multi-dimensional dust environment heat dissipation feature vector into a gradient boosting learning model with a multi-attraction mechanism for processing, so as to obtain fan cooperative control parameters adapted to the dust environment.

[0007] In the technical solution provided by the present invention, a comprehensive dust environment sample library is established through accurate multi-condition data collection and processing. The dust accumulation dynamic characteristics and air duct blockage formation mechanism model developed based on this library reveal the inherent mechanism of reduced heat dissipation efficiency in dust environments. With the help of innovative features such as the sum of periodic temperature differences and energy consumption fluctuation arrays, the system comprehensively characterizes the working state of the heat dissipation system in dust environments, overcoming the inadequate description of traditional features. The constructed multi-attraction mechanism gradient boosting learning model realizes the accurate prediction of the optimal control parameters and has achieved significant improvement in the generation of fan control trajectories. The control strategy combining complementary fan control with dynamic scaling enables the multi-fan system to form a complementary effect to resist dust by reasonably setting the fan phase difference and dynamically adjusting the power distribution, and at the same time realizes an adaptive control parameter adjustment mechanism based on temperature, dust concentration and system reliability index. Compared with the prior art, the present invention improves the control efficiency by an order of magnitude in dust environments, reduces the system energy consumption, and prolongs the service life of the fan, providing an economical and efficient solution for heat dissipation of industrial equipment in dust environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0009] Figure 1 A schematic diagram of the steps of an intelligent adjustment method of an intelligent control system for a heat dissipation fan in an anti-sand and dust environment according to an embodiment of the present invention; Figure 2It is a structural schematic diagram of an intelligent adjustment device of an intelligent control system for a heat dissipation fan in an anti-dust environment according to an embodiment of the present invention. DETAILED DESCRIPTION

[0010] The embodiment of the present invention provides an intelligent adjustment method and device for an intelligent control system of a cooling fan in an anti-dust environment. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0011] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 An embodiment of an intelligent adjustment method of an intelligent control system for a heat dissipation fan in an anti-sand and dust environment according to an embodiment of the present invention includes: Step S1, collecting multi-operating condition data on a heat dissipation test platform of a heat dissipation fan to obtain a dust environment sample library; It is understandable that the execution subject of the present invention may be an intelligent control system intelligent adjustment device for cooling fans in dusty environments, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0012] Specifically, a heat dissipation test platform for the cooling fan is built. The test platform includes a cooling fan system and is equipped with a series of sensor devices for collecting the environment and equipment operation status, including a temperature sensor array, a dust concentration sensor, a fan speed measurement device, a power monitoring module, and a thermal imager. The temperature sensor array monitors the temperature distribution changes in the heat dissipation area in real time, the dust concentration sensor is used to measure the dust concentration level in the test environment, and the fan speed measurement device records the speed data of the fan under different working conditions. The power monitoring module measures the power consumption performance of the fan system, and the thermal imager can intuitively present the heat distribution of the cooling fan under different load and environmental conditions. The temperature sensor array is configured to be evenly distributed at the key target positions in the heat dissipation area to avoid blind spots and errors in temperature data collection caused by uneven distribution of sensors. At the same time, the sampling frequency of the sensor is set so that data collection can ensure real-time performance without causing data redundancy and waste of system resources due to excessively high sampling frequency. After completing the hardware configuration, the test environment is set up with multiple working conditions, and a test matrix containing multi-dimensional parameters is formulated. The multi-dimensional test matrix includes M dust concentration levels, N ambient temperature ranges, and T heat load levels. Through the multi-dimensional combined test method, various dust environments and operating conditions faced by the cooling fan are covered, thereby ensuring the representativeness and effectiveness of the test data. Based on the multi-dimensional test matrix, the single fan mode and the multi-fan mode are monitored, so that the cooling effect and system performance under different fan configurations can be compared. During the test, the temperature change data is obtained in real time through the temperature sensor array, the dust concentration sensor continuously records the dust accumulation data, the fan speed measurement device synchronously obtains the fan speed data under different loads, and the power monitoring module monitors the energy consumption changes of the fan system. Outlier detection and filtering are performed on the temperature change data, dust accumulation data, fan speed data, and power consumption data. The outlier detection process uses a combination of statistical methods and machine learning models to effectively eliminate data anomalies caused by sensor errors, environmental emergencies, or equipment failures. For example, by setting a threshold range based on historical data and physical models, we can quickly filter out abnormal data points that deviate from the normal range, and use filtering processing (such as Kalman filtering, mean filtering, etc.) to smooth the data curve and eliminate noise interference to obtain a valid data set after denoising. The denoised valid data set is classified and labeled according to dust concentration, ambient temperature and heat load. According to different dust concentration levels (such as M concentration levels from low to high), different ambient temperature ranges (N temperature ranges, such as low temperature, medium temperature and high temperature environment) and different heat load levels (T heat load levels, such as no load, partial load and full load), all data are orderly organized, classified and labeled to obtain a dust environment sample library.

[0013] Step S2: quantitatively modeling and analyzing the dust accumulation dynamics characteristics and the air duct blockage formation mechanism based on the dust environment sample library to obtain a thermal equivalent model; Specifically, we conducted correlation analysis on the dust accumulation data in the dust environment sample library and the operating parameters of the cooling fan. By analyzing the accumulation law of dust in the cooling fan system under different working conditions, we constructed a dust accumulation dynamics model. In this process, we established a mathematical model of dust accumulation based on the dynamic characteristics of the accumulation over time. For example, we used the accumulation rate function to describe the influence of dust concentration, wind speed, temperature and other factors on the dust accumulation process. Assuming that the dust accumulation amount Over time The change process is affected by the dust concentration , wind speed and ambient temperature The dynamic model is established as ,in is the initial dust amount, is a cumulative rate function obtained through fitting analysis of sample library data, which accurately describes the dynamic process of dust accumulation on fan blades and duct surfaces. In order to analyze the formation mechanism of duct blockage, the heat dissipation area is meshed, and the entire heat dissipation duct and related heat conduction areas are divided into several small grids. By calculating the thermal resistance of each grid point, a refined thermal resistance distribution grid is obtained. Meshing can not only accurately locate the blocked area in the duct in space, but also deduce the impact of blockage on the overall airflow distribution and heat dissipation efficiency of the duct through the thermal resistance calculation model. For example, at each grid point On, thermal resistance The comprehensive calculation of material thermal conductivity, dust accumulation thickness and wind speed is used to construct the principle model of air duct blockage. The air duct blockage model is further quantified as ,in is the initial thermal resistance, is the coefficient of influence of dust accumulation on the increase of thermal resistance. This model accurately reflects the dynamic influence of dust accumulation on the duct resistance and airflow distribution of the cooling fan. The actual and theoretical cooling efficiencies of the cooling fan under different working conditions are compared and analyzed to obtain the fan performance attenuation model. During the experiment, the fan's cooling performance data is collected in real time through the thermal imager, power monitoring module and temperature sensor array, and the actual measured cooling efficiency is converted into Compared with the theoretical heat dissipation efficiency calculated under ideal conditions By comparison, it is found that the heat dissipation efficiency decreases due to dust accumulation and air duct blockage. Based on these data, a fan performance attenuation model is constructed, such as through the efficiency attenuation coefficient To describe the effects of dust concentration, thermal resistance change and fan speed change on heat dissipation performance. Perform parameter fitting analysis on the dust accumulation dynamics characteristic model, the air duct blockage formation principle model and the fan performance attenuation model. Through optimization methods such as least squares method, gradient descent method or genetic algorithm, fit and compare the experimental data with the model output, and adjust the parameters in the model so that the model can reflect the actual situation to the greatest extent. For example, the dust adhesion rate coefficient is calculated through the experimental data of dust accumulation and concentration, wind speed, temperature and other variables. and dust concentration influence coefficient The dust accumulation model More accurate. The thermal resistance influence coefficient is fitted through the relationship between the thermal resistance change and the dust accumulation in the air duct blockage model. In order to more accurately predict the impact of air duct blockage on heat dissipation performance, the optimal parameter values ​​obtained by fitting are respectively substituted into the dust accumulation dynamic characteristic model, the air duct blockage formation principle model and the fan performance attenuation model, and the final thermal equivalent model is formed through the combination of models.

[0014] Step S3, extracting and integrating features based on the thermal equivalent model to obtain a multi-dimensional dust environment heat dissipation feature vector; Specifically, the data output by the thermal equivalent model is preliminarily screened to construct the original feature set. The thermal equivalent model provides the operating data of the cooling fan system in different dusty environments, including instantaneous temperature value, temperature change rate, dust concentration, fan speed and power consumption data. These data are collected in real time by sensors and stored in the data set after preprocessing. By analyzing and screening these data, outliers and invalid data are eliminated, and the key features that effectively reflect the system status under normal working conditions are retained to obtain the original feature set. The temperature data in the original feature set is segmented, and the entire temperature data time series is divided into several time windows of fixed length (such as one window per minute or hour). The difference between the maximum and minimum temperature values ​​is calculated in each time window, and these differences are accumulated to obtain the sum feature of periodic temperature differences, which effectively captures the fluctuation of the cooling capacity of the cooling fan in different time periods. This feature is also used to detect the temperature stability of the cooling fan at night or under low load conditions, so as to assist in determining whether the system has entered a stable operation mode. At the same time, the power data in the original feature set is analyzed by continuous recording, the standard deviation of the power data in each time period is calculated, and these standard deviation values ​​are arranged into a sequence to obtain the energy consumption fluctuation array feature. The standard deviation of the power data can quantify the fluctuation amplitude of the energy consumption of the fan system in a specific time period. When the fan is in a stable operation state, its power fluctuation is small, and when the system is blocked by dust or the load changes drastically, the power fluctuation will increase significantly. The fan speed data in the original feature set is analyzed by Fourier transform. By converting the time domain signal into a frequency domain signal, the energy distribution in the speed spectrum is calculated, and the ratio of the spectrum energy to the energy of the original speed signal is calculated to obtain the speed regularity feature, which reflects whether the fan maintains a certain operating rhythm during long-term operation. For example, under normal circumstances, the fan speed should show a certain regular fluctuation with the change of temperature or load, and when the fan speed frequently jumps due to dust blockage, its spectrum energy ratio will change significantly. In the analysis of the heat dissipation efficiency data, the number of times the heat dissipation efficiency change direction is reversed is counted by the difference and sign change detection method, and the heat dissipation efficiency change reversal number feature is obtained. This feature is used to capture the stability and response characteristics of the system's heat dissipation efficiency under different working conditions. The original feature set, the sum of periodic temperature differences, the energy consumption fluctuation array feature, the speed regularity feature, and the number of reversal times of heat dissipation efficiency changes are combined to form an initial multi-dimensional feature vector, and feature dimensionality reduction processing is performed. For example, the high-dimensional feature vector is projected into a low-dimensional space through principal component analysis, linear discriminant analysis, or autoencoder, while trying to maintain the effective information of the feature. In the process of dimensionality reduction, the key features that have a greater impact on the heat dissipation performance are automatically selected, and the redundant and noise information is eliminated to obtain a multi-dimensional dust environment heat dissipation feature vector.

[0015] Step S4: input the multi-dimensional dust environment heat dissipation feature vector into the gradient boosting learning model with a multi-attraction mechanism for processing, and obtain the fan cooperative control parameters adapted to the dust environment.

[0016] Specifically, the multi-dimensional dust environment heat dissipation feature vector is input into the deep neural network for feature mapping to obtain the mapping feature space data. The deep neural network performs multiple nonlinear transformations on the feature data through its multi-layer structure and nonlinear activation function, so that the feature vector completes complex feature mapping in the high-dimensional space, thereby converting the input features into high-level feature representations with more predictive value. In this process, the hidden layer of the network gradually extracts the patterns and rules in the data, so that the mapping feature space data is not only a simple transformation of the original data, but also a deep reconstruction at the data abstraction level. The mapping feature space data is input into the gradient boosting learning model with a multi-attraction mechanism for control parameter prediction to obtain the preliminary prediction value of the control parameter. The core module of the gradient boosting learning model consists of a gradient boosting decision module constructed by the XGBoost algorithm. XGBoost is an ensemble learning method based on decision trees. By constructing multiple weak classifiers (decision trees) to gradually fit the data errors, the model can continuously learn the nonlinear relationship between features and target outputs. In the specific implementation, the model generates a new decision tree to supplement and correct the prediction error of the previous round through the gradient information of the error in each round of iteration learning. This mechanism of gradually approaching the true value can effectively improve the prediction accuracy of the model. The regularization mechanism in the XGBoost model can also avoid overfitting of the model, so that the initial predicted values ​​of the predicted control parameters can maintain good generalization performance in various dust environments. The similarity between the heat dissipation feature vector of the multidimensional dust environment and the predefined K attractors is calculated to obtain the similarity value between the feature vector and each attractor. Attractors refer to predefined feature patterns under different dust environments and working conditions, such as specific attractors for high dust concentration and low wind speed environments, or attractors for high temperature and high load conditions. These attractors are predefined through cluster analysis or expert experience, representing the control requirements of the cooling system under typical environments. By calculating the similarity between the current feature vector and these attractors, the matching degree between the current working condition and the historical typical working condition is quantified, for example, the similarity value is calculated by cosine similarity, Euclidean distance or dynamic time warping method. Based on the similarity value between the feature vector and each attractor, normalization is performed to obtain the weight coefficient to ensure that the influence weights of different attractors remain stable and comparable in value. Normalization processing uses a soft maximum function or linear normalization method to convert all similarity values ​​into a probability distribution between 0 and 1, so that the sum of all weight coefficients is 1, effectively avoiding weight imbalance caused by individual similarity values ​​being too large or too small, thereby making the control parameter prediction process more stable and reliable. The preliminary predicted values ​​of the control parameters corresponding to each attractor are weighted and fused according to the weight coefficient to obtain the fan collaborative control parameters. The preliminary predicted value corresponding to each attractor is multiplied by its weight coefficient, and then the weighted predicted values ​​of all attractors are accumulated to obtain the weighted fused control parameters.

[0017] The historical optimization control strategy data are clustered to obtain K typical control strategy center points, which are used as K attractors. Each attractor represents an optimal control mode in the feature space. The cluster analysis method uses the K-means clustering algorithm, which divides the historical data into K clusters in the feature space through repeated iterations. The center of each cluster is an attractor. The historical optimization control strategy data includes the fan control parameters (such as speed, power allocation ratio, response time threshold) and corresponding system feedback (such as heat dissipation efficiency, energy consumption, temperature stability, etc.) under different dust environments and fan operating conditions. Through cluster analysis, these data are aggregated into K representative control modes, so that each attractor can reflect the optimal control strategy under specific working conditions. In the subsequent control process, when the actual feature vector has a high similarity with a certain attractor, it means that the current working condition is highly matched with the optimal control mode represented by the attractor, and the system gives priority to the control parameters of this mode. The multi-dimensional dust environment heat dissipation feature vector and K attractors are normalized, and the standardization method (z-score standardization) method is used to map the feature data to a specific numerical range (such as between 0 and 1), so that the contribution of the features of each dimension to the similarity calculation will not be biased due to different dimensions. For example, for features such as temperature, fan speed, dust concentration, and power consumption, the problem of different original numerical ranges of these features is eliminated through normalization, so that the feature vector and attractor can perform accurate distance calculation in the same standardized space. The Euclidean distance between the standardized feature vector and each standardized attractor is calculated to obtain K distance values, and the similarity between the current feature vector and each attractor is quantitatively evaluated. The smaller the distance, the higher the similarity. The distance value is transformed by the Gaussian kernel function to obtain a smoother initial similarity value with an upper limit of 1. Nonlinear transformation can effectively avoid the problem of similarity approaching 0 when the distance value is too large, thereby maintaining the system's responsiveness to medium and long-distance attractors, making the weight distribution of each attractor in the multi-attraction mechanism smoother and more continuous in the decision-making process. The initial similarity value is dynamically adjusted based on the dust concentration and load level to obtain the similarity value between the feature vector and each attractor. The priority of different attractors is dynamically adjusted according to the current environmental characteristics. For example, in an environment with high dust concentration, the system is more inclined to select attractors that perform well in similar high dust environments. The dust concentration is associated with the attractor preference by applying a weight adjustment function to the initial similarity value. For example, the similarity value is re-weighted by a linear function or an exponential function, so that when the dust concentration is high, the similarity value of the attractor in the high dust environment is amplified, while the similarity value of the attractor in the low dust environment is moderately weakened. Similarly, when the load level is high, through a similar method, the system is more inclined to select attractors that can maintain a balance between heat dissipation efficiency and energy consumption in a high load environment, thereby ensuring the rationality and safety of the fan control parameters.

[0018] In a multi-fan system, in order to ensure that each fan works in coordination and avoids airflow interference, the phase of each fan in the system is defined to obtain the fan control phase parameter. In a multi-fan system, each fan is assigned a specific phase angle so that the phase difference between adjacent fans is maintained at 2π / N (where N is the total number of fans). By setting the evenly distributed phase difference, the airflow generated by the fan during operation forms periodic fluctuations, avoiding the problem of excessive concentration or turbulence of airflow caused by the synchronous operation of multiple fans, and achieving a more stable heat dissipation airflow distribution. Based on the defined fan control phase parameters, the basic speed of each fan is calculated in combination with the speed adjustment curve in the fan cooperative control parameters to obtain the speed reference value. According to the current dust environment characteristics, dust concentration, temperature distribution and load level, the intelligent control model is used to predict a speed adjustment curve that adapts to the current working conditions. The curve defines the optimal speed range of the fan under different time or temperature conditions. After obtaining the speed reference value, the speed of each fan is sinusoidally modulated according to the fan control phase parameters, so that the fan speed can not only be kept within the optimal adjustment curve, but also the alternating working state of each fan can be realized through the periodic characteristics of the sine function. Through this sinusoidal modulation method, each fan always maintains a certain phase difference during operation, thereby achieving the effect of alternating airflow output, enhancing the uniformity of the heat dissipation airflow, and effectively reducing the mechanical stress and current shock caused by the synchronous start and stop of the fan, thereby extending the service life of the fan. In order to enable the fan system to dynamically adjust the power allocation ratio according to the actual heat dissipation demand, the power allocation ratio in the fan cooperative control parameters is dynamically adjusted according to the temperature difference in the area responsible for each fan, and a temperature-adaptive power allocation ratio is obtained. The temperature data around each fan is monitored in real time by a temperature sensor, and these temperature data are compared with the actual heat dissipation effect of the fan. When the temperature in the area responsible for a certain fan is significantly higher than that in other areas, the system automatically increases the power allocation ratio of the fan, thereby improving the heat dissipation intensity of the area. At the same time, in a dusty environment, since dust accumulation will affect the heat dissipation efficiency and mechanical properties of the fan, the dust-adaptive response time is obtained by calculating the response time threshold in the fan cooperative control parameters combined with the dust accumulation of each fan. The dust accumulation around each fan is monitored in real time through dust sensors. When the dust accumulation reaches a certain threshold, the fan response time is automatically shortened, the fan speed is increased to enhance the self-cleaning ability, or the anti-air blowing mechanism is activated to remove the dust attached to the fan blades, thereby delaying the decay rate of the fan performance. In order to cope with sudden changes in dust concentration, the rate of change of dust concentration in the environment is monitored in real time. When a significant sudden change in dust concentration is detected, the emergency adjustment mechanism is immediately triggered to obtain a speed curve optimized for dust prevention.In this case, the system will temporarily increase the speed of all fans to form a stronger airflow resistance to prevent dust from quickly entering the equipment. At the same time, the fan operation mode will be adjusted according to the air duct conditions, such as switching to negative pressure mode to enhance dust removal. The temperature-adaptive power allocation ratio, dust-adaptive response time, and dust-optimized speed curve are combined and applied to the control of each fan to obtain a complementary fan control strategy. Under this strategy, each fan achieves alternating airflow output through dynamic phase regulation, and the dual mechanisms of temperature adaptation and dust adaptation ensure the long-term stable operation of the system in complex dust environments. The emergency adjustment mechanism is used to respond to extreme dust events, thereby achieving intelligent, automated, and efficient fan control effects.

[0019] In the embodiment of the present invention, a comprehensive dust environment sample library is established through accurate multi-condition data collection and processing. The dust accumulation dynamic characteristics and air duct blockage formation mechanism model developed based on this library reveal the internal mechanism of reduced heat dissipation efficiency in dust environments. With the help of innovative features such as the sum of periodic temperature differences and energy consumption fluctuation arrays, the system comprehensively characterizes the working state of the heat dissipation system in dust environments, overcoming the inadequate description of traditional features. The constructed multi-attraction mechanism gradient boosting learning model realizes the accurate prediction of the optimal control parameters and has achieved significant improvement in fan control trajectory generation. The control strategy combining complementary fan control with dynamic scaling enables the multi-fan system to form a complementary effect to resist dust by reasonably setting the fan phase difference and dynamically adjusting the power distribution, and at the same time realizes an adaptive control parameter adjustment mechanism based on temperature, dust concentration and system reliability index. Compared with the prior art, the present invention improves the control efficiency by an order of magnitude in dust environments, reduces the system energy consumption, and prolongs the service life of the fan, providing an economical and efficient solution for heat dissipation of industrial equipment in dust environments.

[0020] In a specific embodiment, the process of executing step S1 may specifically include the following steps: A heat dissipation test platform for the cooling fan is built to obtain a test environment including a cooling fan system, a temperature sensor array, a dust concentration sensor, a fan speed measurement device, a power monitoring module and a thermal imager; Configure the temperature sensor array to be evenly distributed at the target position of the heat dissipation area and set the sampling frequency of the sensor. At the same time, set the working conditions of the test environment to obtain a multi-dimensional test matrix containing M dust concentration levels, N ambient temperature ranges and T heat load levels; Based on the multi-dimensional test matrix, the single-fan mode and multi-fan mode are monitored to obtain temperature change data, dust accumulation data, fan speed data and power consumption data; The temperature change data, dust accumulation data, fan speed data and power consumption data are subjected to outlier detection and filtering to obtain a valid data set after denoising. The valid data set after denoising is then classified and labeled according to dust concentration, ambient temperature and heat load to obtain a dust environment sample library.

[0021] Specifically, a complete test environment is designed, which includes a cooling fan system and is equipped with a series of professional equipment for monitoring environmental parameters and fan operation status, including temperature sensor arrays, dust concentration sensors, fan speed measurement devices, power monitoring modules, and thermal imagers. In the process of building the test platform, a fan system suitable for the test space is selected, and its heat dissipation capacity, wind speed range, and speed adjustment capacity must be able to cover the needs under different working conditions. For example, a cooling fan with variable speed, high air volume, and PWM (pulse width modulation) control is selected to ensure that the fan speed can be dynamically adjusted in a simulated dust environment to achieve switching of multiple cooling modes. The temperature sensor array is one of the core components used in the test platform to monitor temperature changes. These sensors need to be evenly distributed at the target position of the heat dissipation area to ensure the comprehensiveness and accuracy of temperature data collection. When arranging sensors, a gridding method is used to divide the entire heat dissipation area into several small areas, and a temperature sensor is arranged in each small area to form a two-dimensional or three-dimensional temperature monitoring grid. At the same time, a suitable sampling frequency is set for the sensor. During the test, the dust concentration sensor is used to detect the change of dust concentration in the environment, the fan speed measuring device obtains the running speed of the fan in real time, the power monitoring module is used to accurately calculate the power consumption performance of the fan system, and the thermal imager can provide an intuitive temperature distribution image. In order to make the test environment closer to the actual working conditions, multi-dimensional working conditions are set to build a multi-dimensional test matrix containing M dust concentration levels, N ambient temperature ranges and T heat load levels. During the experiment, the system is monitored in single fan mode and multi-fan mode to compare the impact of different fan configurations on heat dissipation performance. For example, in single fan mode, by adjusting the speed and wind direction of the fan, the temperature change and heat dissipation effect under different dust concentrations and heat loads are observed. In multi-fan mode, by adjusting the phase difference and speed synchronization of the fan, the airflow distribution and heat dissipation efficiency differences when the fans work together are analyzed. During the test, the temperature sensor array continuously records the temperature change data of the heat dissipation area, the dust concentration sensor monitors the dust accumulation in the environment in real time, the fan speed measuring device obtains the running speed data of the fan, the power monitoring module records the power consumption data of the fan system, and the thermal imager provides a visual image of the temperature distribution. These data are stored in real time by the data acquisition system to form a preliminary data set. The temperature change data, dust accumulation data, fan speed data, and power consumption data are subjected to outlier detection and filtering to obtain a valid data set after denoising. Outlier detection uses statistical methods, such as methods based on standard deviation or interquartile range to identify outliers in the data and mark data points that are outside the normal range as outliers. For example, for temperature data, if the temperature reading of a sensor exceeds three standard deviations of the mean, it is considered an outlier and removed from the data set.The filtering process uses Kalman filtering or moving average filtering to eliminate noise interference by smoothing the data curve. For example, for fan speed data, the moving average filter eliminates the reading jitter caused by the instantaneous fluctuation of the fan, and obtains a more stable speed data sequence. The effective data set after denoising is classified and labeled according to dust concentration, ambient temperature and heat load to build a dust environment sample library. Each record in the sample library includes raw data such as temperature, dust, speed and power consumption, and is accompanied by a clear description of the working conditions.

[0022] In a specific embodiment, the process of executing step S2 may specifically include the following steps: The dust accumulation data and operating parameters in the dust environment sample library were correlated and analyzed to obtain the dust accumulation dynamics characteristic model; The heat dissipation area is meshed to obtain a thermal resistance distribution grid, and the thermal resistance is calculated for each grid point of the thermal resistance distribution grid to obtain a principle model of air duct blockage formation; Compare and analyze the actual and theoretical cooling efficiencies of cooling fans under different working conditions to obtain a fan performance attenuation model. The parameter fitting analysis was performed on the dust accumulation dynamics characteristic model, the air duct blockage formation principle model and the fan performance attenuation model to obtain the optimal parameter values ​​including the dust adhesion rate coefficient, the dust concentration influence coefficient, the dust direct influence coefficient and the thermal resistance influence coefficient. The optimal parameter values ​​are substituted into the dust accumulation dynamics characteristic model, the air duct blockage formation principle model and the fan performance attenuation model respectively to obtain a thermal equivalent model.

[0023] Specifically, the dust accumulation data and related operating parameters, including fan speed, wind speed, dust concentration, temperature and humidity, are extracted from the dust environment sample library, and the dynamic relationship between these parameters and dust accumulation characteristics is revealed through correlation analysis. In the process of modeling the dynamic characteristics of dust accumulation, a dynamic model is constructed by analyzing the law of dust accumulation over time. This model not only considers the influence of dust concentration and wind speed on dust deposition rate, but also introduces ambient temperature and humidity as correction factors. The dust accumulation dynamic model is expressed as:

[0024] in, Indicates time The amount of dust accumulated at a given moment, is the initial dust amount, is the dust concentration, is the wind speed, is the ambient temperature, is humidity, and are the dust adhesion rate coefficient, temperature influence coefficient and humidity attenuation coefficient respectively. In this model, the first term It reflects the direct influence of dust concentration and wind speed on dust adhesion speed, while the second The inhibitory effect of humidity on dust adhesion is described in the form of exponential decay. For example, in a high humidity environment, dust is easy to clump and is not easy to adhere to the fan blades, thereby reducing the accumulation rate. The heat dissipation area is meshed to construct a thermal resistance distribution grid, and the thermal resistance of each grid point is calculated to obtain the principle model of air duct blockage. The entire heat dissipation duct area is divided into The thermal resistance value at each grid point is determined by the dust accumulation and air flow velocity. The thermal resistance calculation model is expressed as:

[0025] in, It is The thermal resistance value of each grid point is is the initial thermal resistance, is the dust accumulation at the grid point, is the air velocity, is the thermal resistance influence coefficient, is a small positive number that prevents the denominator from being zero. In this model, the dust accumulation The increase in thermal resistance The air flow rate increases linearly The improvement can effectively reduce the local thermal resistance. This model can reflect the dynamic process of dust accumulation blocking the heat dissipation air duct, and can also reveal the effect of wind speed optimization in improving the air duct resistance. In order to evaluate the heat dissipation performance of the fan under different working conditions, the actual heat dissipation efficiency and the theoretical heat dissipation efficiency are compared and analyzed to obtain the fan performance attenuation model. The actual heat dissipation efficiency is obtained through the temperature sensor and power monitoring module of the test platform, while the theoretical heat dissipation efficiency is calculated based on the ideal operation model of the fan. For example, under ideal conditions without dust blockage and air duct interference, the heat dissipation power of the fan is It is expressed as:

[0026] in, is the specific heat capacity of air, is the air volume, is the temperature difference between the inlet and outlet. The actual heat dissipation efficiency Due to the influence of dust accumulation and air duct blockage, it will be lower than the theoretical value, so the fan performance attenuation model is expressed as:

[0027] in, is the heat dissipation efficiency attenuation coefficient, and They are the dust direct influence coefficient and thermal resistance influence coefficient respectively. and Respectively represent the overall dust accumulation and average thermal resistance of the system. In this model, by introducing the dual influencing factors of dust accumulation and thermal resistance change, the attenuation of fan performance over time can be more accurately evaluated. The parameter fitting analysis of the dust accumulation dynamics characteristic model, the air duct blockage formation principle model and the fan performance attenuation model is carried out, and the optimal parameter values ​​of the dust adhesion rate coefficient, dust concentration influence coefficient, dust direct influence coefficient and thermal resistance influence coefficient are obtained by fitting the experimental data by the least squares method. This process is achieved by minimizing the error between the model prediction value and the actual measurement value. The specific objective function is expressed as:

[0028] in, is the experimentally measured heat dissipation efficiency attenuation value, is the heat dissipation efficiency predicted by the model, and by optimizing the model parameters (i=1,2,…,6), so that the prediction results are as close to the experimental data as possible, thus ensuring the accuracy and reliability of the model in practical applications. The optimal parameter values ​​are substituted into the dust accumulation dynamics model, the air duct blockage formation principle model and the fan performance attenuation model to obtain a thermal equivalent model, which comprehensively predicts the cooling efficiency and operating status of the cooling fan in different dust environments. For example, when a specific dust concentration, wind speed, temperature and humidity parameters are input, the thermal equivalent model can output the fan's expected cooling power, power consumption and cooling efficiency change trend.

[0029] In a specific embodiment, the process of executing step S3 may specifically include the following steps: The instantaneous temperature value, temperature change rate, dust concentration, fan speed and power consumption data output from the thermal equivalent model are screened to obtain the original feature set; The temperature data in the original feature set is processed in segments, and the cumulative sum of the difference between the maximum and minimum temperature in each time window is calculated to obtain the periodic temperature difference sum feature; The power data in the original feature set is continuously recorded, and the sequence composed of the standard deviation of the power in each time period is calculated to obtain the energy consumption fluctuation array feature; Perform Fourier transform analysis on the fan speed data in the original feature set, calculate the ratio of the speed spectrum energy to the original speed signal energy, and obtain the speed regularity feature; Perform differential and sign change detection on the heat dissipation efficiency data in the original feature set, count the number of times the heat dissipation efficiency change direction is reversed, and obtain the heat dissipation efficiency change reversal number feature; The original feature set, the sum feature of periodic temperature difference, the energy consumption fluctuation array feature, the rotation speed regularity feature and the number of reversal times of heat dissipation efficiency change are combined and reduced in dimension to obtain a multi-dimensional dust environment heat dissipation feature vector.

[0030] Specifically, among the various data output by the thermal equivalent model, the instantaneous temperature value, temperature change rate, dust concentration, fan speed and power consumption data are screened through a preprocessing algorithm to obtain an original feature set. In this process, some noise data and outliers are eliminated. For example, the temperature and speed data are smoothed by moving average filtering or Kalman filtering. At the same time, the triple standard deviation method (3σ criterion) is used to detect and eliminate outliers to ensure the data quality in the original feature set. When constructing the original feature set, all data points are timestamped to ensure that the changing trend of the data in the time dimension can be accurately located in the subsequent feature extraction process. After obtaining the original feature set, the periodic temperature difference sum feature is extracted for the temperature data through segmentation processing. The temperature data is segmented according to fixed time windows, and the maximum temperature is calculated in each time window. and minimum value , and then calculate the cumulative sum of these temperature differences:

[0031] in, represents the periodic temperature difference sum characteristic, is the number of time windows, and Respectively The maximum and minimum temperature values ​​in a time window. This feature can reflect the temperature fluctuations of the cooling system under different loads and environmental conditions. Through this feature value, it is judged whether the cooling capacity of the current fan matches the actual heat load, so as to adjust the fan speed or power distribution when necessary. At the same time, for the power data in the original feature set, the energy consumption fluctuation array feature is obtained by calculating the standard deviation sequence of power in each time period. The power data is segmented according to the same time window, and the standard deviation of the power data is calculated in each time window , and then group these standard deviation values ​​into a sequence: .in, is the energy consumption fluctuation array characteristic, Indicates The standard deviation of the power data within a time window. This feature can quantify the fluctuation of the power consumption of the cooling fan under different working conditions. When analyzing the fan speed data, the speed signal is analyzed in the frequency domain through Fourier transform to extract the speed regularity characteristics. Perform Fourier transform to get the spectrum , and then calculate the ratio of the spectrum energy to the original signal energy:

[0032] in, Indicates the regularity characteristics of the speed. and It is the characteristic frequency range set by the system. The spectrum energy within this range indicates the regular changes of the fan at a specific frequency. For example, in the alternating speed regulation or periodic dust cleaning mode, the speed signal will have a strong harmonic component at a specific frequency. By calculating this ratio, the periodicity and stability of the fan operation are quantified, so that the system can monitor in real time whether the fan is in the preset operation mode. For the heat dissipation efficiency data, the number of times the direction of the heat dissipation efficiency change is reversed is counted through the difference and sign change detection method to obtain the heat dissipation efficiency change reversal number characteristics. Calculate the heat dissipation efficiency data The difference sequence of:

[0033] Then count the number of times the sign changes in the differential sequence, that is, the total number of times it changes from positive to negative or from negative to positive, and record it as . This feature reflects the stability of the heat dissipation efficiency. When the system is in a stable state, the change in heat dissipation efficiency should be smooth and consistent in direction. When the load changes frequently or the air duct is blocked, the heat dissipation efficiency will fluctuate violently, resulting in a significant increase in the number of sign changes. The original feature set, the sum of periodic temperature differences, the energy consumption fluctuation array feature, the speed regularity feature, and the heat dissipation efficiency change reversal number feature are combined to obtain the initial high-dimensional feature vector. Through dimensionality reduction processing methods, such as principal component analysis, the high-dimensional feature vector is reduced to a more compact form, and finally a multi-dimensional dust environment heat dissipation feature vector is obtained. .in, is the feature dimension after dimensionality reduction, Represents each important feature component.

[0034] In a specific embodiment, the process of executing step S4 may specifically include the following steps: The multi-dimensional dust environment heat dissipation feature vector is input into the deep neural network for feature mapping to obtain mapping feature space data; Inputting the mapped feature space data into a gradient boosting learning model with a multi-attraction mechanism to predict the control parameters, and obtaining a preliminary prediction value of the control parameters. The gradient boosting learning model with a multi-attraction mechanism includes a gradient boosting decision module constructed by an XGBoost algorithm; The similarity between the heat dissipation feature vector of the multi-dimensional dust environment and the predefined K attractors is calculated to obtain the similarity value between the feature vector and each attractor; Based on the similarity value between the eigenvector and each attractor, normalization is performed to obtain the weight coefficient; The preliminary predicted values ​​of the control parameters corresponding to each attractor are weighted and fused according to the weight coefficient to obtain the fan collaborative control parameters. The collaborative control parameters include the speed regulation curve, the power allocation ratio and the response time threshold.

[0035] Specifically, the multi-dimensional dust environment heat dissipation feature vector Input into a trained deep neural network, and through multi-layer nonlinear transformation, map the input feature vector into a high-dimensional feature space to obtain the mapped feature space data In the feature mapping process, each layer of the deep neural network introduces nonlinear characteristics through the activation function, enabling the model to capture the complex relationships in the feature vector. For example, the mapping features are calculated through the fully connected layer in the neural network:

[0036] in, It is mapping features, is the weight coefficient, is the bias, is an activation function (such as ReLU or Sigmoid). After calculations in a multi-layer network, high-dimensional, nonlinearly transformed feature data is obtained. . Map feature space data The input is sent to the gradient boosting learning model with a multi-attraction mechanism. The core of the model is the gradient boosting decision module built based on the XGBoost algorithm. XGBoost builds multiple decision tree models and uses the prediction error of the previous round for learning in each iteration, thereby gradually improving the prediction accuracy of the model. In this process, the output of each tree is a fit to the residual of the previous round of prediction, that is, the model parameters are optimized by minimizing the objective function:

[0037] in, is the objective function, is the sample size, is the loss function, which represents the actual value With the predicted value The error between is the number of decision trees, is the model complexity penalty term, It is In practical applications, the XGBoost model generates a new decision tree in each iteration, and gradually corrects the prediction results through the additive model to obtain the preliminary prediction values ​​of the control parameters. These predictions include the speed regulation curve, power allocation ratio and preliminary estimation of response time threshold. After obtaining the preliminary prediction values ​​of the control parameters, in order to introduce the multi-attraction mechanism, the multi-dimensional dust environment heat dissipation feature vector is With predefined Attractor Perform similarity calculation to obtain the similarity value between the feature vector and each attractor Attractors represent typical control modes under different dust environments and working conditions, which are predefined through cluster analysis or expert experience. For example, each attractor represents an optimal fan control strategy. In the similarity calculation, a similarity measurement method based on Gaussian kernel function is used, for example, by calculating the Euclidean distance between the feature vector and each attractor , and convert the distance value to a similarity value:

[0038] in, is the eigenvector and The similarity value of attractors, is the bandwidth parameter of the Gaussian kernel function, is an attractor Middle In this way, the distance between the feature vector and the attractor is mapped to a similarity value. The smaller the distance, the higher the similarity value. Normalize and get the weight coefficient , the weight coefficient represents the contribution of each attractor in the final control parameter. Normalization is achieved through the Softmax function:

[0039] in, It is The weight coefficients of attractors are calculated by the Softmax function, which can convert all similarity values ​​into probability distributions between 0 and 1, so that the sum of all weight coefficients is 1, thereby ensuring the numerical stability and interpretability of weighted fusion calculations. According to these weight coefficients, the preliminary predicted values ​​of the control parameters corresponding to each attractor are weighted and fused to obtain the fan collaborative control parameters. This process is achieved by calculating the weighted sum:

[0040] in, Indicates the final fan collaborative control parameters, including speed adjustment curve, power allocation ratio and response time threshold. It is The initial predicted values ​​of the control parameters corresponding to the attractors are obtained. Through weighted fusion calculation, the predicted values ​​of each attractor are integrated according to the degree of similarity. The final control parameters can reflect the actual situation of the current dust environment, inherit the optimal control experience in historical working conditions, and realize dynamic and intelligent fan control.

[0041] In a specific embodiment, the execution step calculates the similarity between the multi-dimensional dust environment heat dissipation feature vector and the predefined K attractors, and the process of obtaining the similarity value between the feature vector and each attractor may specifically include the following steps: Perform cluster analysis on historical optimization control strategy data to obtain K typical control strategy center points, which are used as K attractors, and each attractor represents an optimal control mode in the feature space; The heat dissipation characteristic vectors and K attractors of the multi-dimensional dust environment are normalized to obtain standardized characteristic vectors and standardized attractors. Calculate the Euclidean distance between the standardized feature vector and each standardized attractor to obtain K distance values, and perform Gaussian kernel function transformation on each distance value to obtain the initial similarity value; The initial similarity value is dynamically adjusted based on the dust concentration and load level to obtain the similarity value between the eigenvector and each attractor.

[0042] Specifically, the control parameters under all historical working conditions are extracted from the historical optimization control strategy data, including fan speed, power allocation ratio, response time threshold, and environment-related features (such as dust concentration, temperature, humidity, and load level). These data points are represented as a multidimensional vector set in the feature space. ,in is the number of samples of historical data, each sample is a In order to identify representative control strategy patterns in these data, a clustering algorithm is used to divide the data into The K-means clustering algorithm achieves the aggregation of data points by minimizing the square distance from the sample to the cluster center. The objective function of the K-means algorithm is:

[0043] in, is the clustering objective function, It is The center point (attractor) of the cluster, Represents data points With cluster center Through iterative calculation, the K-means algorithm eventually converges to a state where the distance from all data points to their cluster centers is minimized, and we get Typical control strategy center point These center points represent typical control modes in historical data and serve as attractors in the multi-attraction mechanism, providing a reference template for the prediction of control parameters of the current system. The attractors are normalized to ensure that the feature data and attractors are calculated on the same numerical scale and the dimensional differences between the feature dimensions are eliminated. After the normalization is completed, the Euclidean distance between the standardized feature vector and each standardized attractor is calculated to obtain distance values. The Euclidean distance calculation formula is:

[0044] in, Represents the normalized feature vector and The distance between attractors, and are the eigenvector and the attractor in the The normalized eigenvalue of dimension, is the dimension of the feature vector. The Euclidean distance can quantify the similarity between the current operating condition characteristics and the historical optimal control mode. The smaller the distance, the higher the matching degree between the current characteristics and the attractor. The distance value is transformed by Gaussian kernel function to obtain the initial similarity value:

[0045] in, It is The initial similarity value of attractors, is the bandwidth parameter of the kernel function, which controls the smoothness of the Gaussian function. The Gaussian kernel function can map the distance value to a similarity value between 0 and 1. When the distance is small, the similarity value is close to 1, and when the distance is large, the similarity value decays rapidly to close to 0. This nonlinear mapping helps to highlight the attractor closest to the current feature, thereby improving the system's responsiveness to the optimal control mode.

[0046] In a specific embodiment, the intelligent adjustment method of the intelligent control system of the cooling fan in the anti-sand and dust environment further includes the following steps: The phase of each fan in the multi-fan system is defined to obtain a fan control phase parameter, where the fan control phase parameter satisfies that the phase difference between adjacent fans is 2π divided by the total number of fans; The basic speed of each fan is calculated based on the speed adjustment curve in the fan cooperative control parameter to obtain the speed reference value, and the speed of each fan is sinusoidally modulated according to the fan control phase parameter to make each fan form an alternating working state; Dynamically adjust the power allocation ratio in the fan cooperative control parameters according to the temperature difference of each fan's responsible area to obtain a temperature-adaptive power allocation ratio; The dust adaptive response time is calculated based on the response time threshold in the fan collaborative control parameter and the dust accumulation of each fan; The rate of change of dust concentration in the environment is monitored in real time. When a sudden change in dust concentration is detected, the emergency adjustment mechanism is triggered to obtain a speed curve optimized for dust prevention. The temperature-adaptive power allocation ratio, dust-adaptive response time and dust-optimized speed curve are combined and applied to the control of each fan to obtain a complementary fan control strategy.

[0047] Specifically, the phase of each fan in the multi-fan system is defined to obtain the fan control phase parameter, so that the phase difference between adjacent fans is ,in is the total number of fans, and the formula is as follows:

[0048] Ensure that the phase difference between each fan is uniform, so that the working status of the fans changes alternately to maximize the heat dissipation efficiency. Based on the speed adjustment curve in the fan collaborative control parameters, calculate the basic speed of each fan Assume that the base speed of the fan is determined by the fan load , Ambient temperature And fan performance parameters The basic speed is calculated as follows:

[0049] in, For the The load of a fan, is the ambient temperature, is the fan performance parameter. Perform sinusoidal modulation on the fan speed to ensure that each fan works alternately. Set the actual fan speed to , then the modulation formula is:

[0050] in, is the operating frequency of the system, is the time variable, For the Through this sinusoidal modulation, each fan will start and stop alternately, optimizing the cooling effect and avoiding interference between fans. , dynamically adjust the power allocation ratio to obtain the temperature-adaptive power allocation ratio . The formula is expressed as:

[0051] in, For the The temperature difference of the area covered by each fan is is a proportional constant. Through this dynamic adjustment, the fan can be adaptively adjusted in different temperature zones to ensure that hotter areas receive more heat dissipation support. Considering the impact of dust accumulation on the fan, based on the amount of dust accumulation , calculate the response time adjustment. Assume the base response time is , the relationship between dust accumulation and response time is expressed as:

[0052] in, As the basic response time, is the dust impact coefficient, For the The dust accumulation of each fan. As the dust accumulates, the response time will gradually increase to ensure that the fan can adjust over a longer period of time to avoid performance degradation. Exceeding the preset threshold When the system triggers the emergency adjustment mechanism, the fan speed curve will be adjusted. The speed curve after dust prevention optimization is:

[0053] in, is the adjustment factor, is the current dust concentration. When the dust concentration exceeds the threshold, the fan will be quickly adjusted through this mechanism to avoid performance degradation caused by dust. All control parameters, including temperature-adaptive power allocation ratio , Dust adaptive response time Speed ​​curve after dust protection optimization , combined with the control applied to each fan, a complementary fan control strategy is obtained.

[0054] The above describes the intelligent adjustment method of the intelligent control system of the cooling fan in the dusty environment in the embodiment of the present invention. The following describes the intelligent adjustment device of the intelligent control system of the cooling fan in the dusty environment in the embodiment of the present invention. Figure 2In one embodiment of the present invention, an intelligent control system intelligent adjustment device for a heat dissipation fan in an anti-sand and dust environment includes: The acquisition module is used to collect data of multiple working conditions on the heat dissipation test platform of the heat dissipation fan to obtain a dust environment sample library; Modeling module, which is used to conduct quantitative modeling and analysis on the dust accumulation dynamics characteristics and the formation mechanism of air duct blockage based on the dust environment sample library to obtain a thermal equivalent model; A feature extraction module is used to extract and integrate features based on a thermal equivalent model to obtain a multi-dimensional dust environment heat dissipation feature vector; The processing module is used to input the multi-dimensional dust environment heat dissipation feature vector into the gradient boosting learning model with multi-attraction mechanism for processing, so as to obtain the fan cooperative control parameters adapted to the dust environment.

[0055] Through the cooperation of the above components, a comprehensive dust environment sample library is established through accurate multi-condition data collection and processing. The dust accumulation dynamic characteristics and air duct blockage formation mechanism model developed based on this library reveal the internal mechanism of reduced heat dissipation efficiency in dust environments. With the help of innovative features such as the sum of periodic temperature differences and energy consumption fluctuation arrays, the system comprehensively characterizes the working state of the heat dissipation system in dust environments, overcoming the inadequate description of traditional features. The constructed multi-attraction mechanism gradient boosting learning model realizes the accurate prediction of the optimal control parameters and has achieved significant improvement in fan control trajectory generation. The control strategy combining complementary fan control with dynamic scaling enables the multi-fan system to form a complementary effect to resist dust by reasonably setting the fan phase difference and dynamically adjusting the power distribution, and at the same time realizes an adaptive control parameter adjustment mechanism based on temperature, dust concentration and system reliability index. Compared with the prior art, the present invention improves the control efficiency by an order of magnitude in dust environments, reduces the system energy consumption, and prolongs the service life of the fan, providing an economical and efficient solution for heat dissipation of industrial equipment in dust environments.

[0056] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0057] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0058] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent adjustment method for an intelligent control system of a cooling fan in an anti-dust environment, characterized in that: The method comprises: Collect data from multiple working conditions on the cooling fan's heat dissipation test platform to obtain a dust environment sample library; Based on the dust environment sample library, the dust accumulation dynamic characteristics and the air duct blockage formation mechanism are quantitatively modeled and analyzed to obtain a thermal equivalent model; Based on the thermal equivalent model, feature extraction and integration are performed to obtain a multi-dimensional dust environment heat dissipation feature vector; The multi-dimensional dust environment heat dissipation feature vector is input into a gradient boosting learning model with a multi-attraction mechanism for processing, so as to obtain fan cooperative control parameters adapted to the dust environment.

2. The intelligent adjustment method of the intelligent control system of the cooling fan in the dust-resistant environment according to claim 1, characterized in that: The heat dissipation test platform of the heat dissipation fan collects data under multiple working conditions to obtain a dust environment sample library, including: A heat dissipation test platform for the cooling fan is built to obtain a test environment including a cooling fan system, a temperature sensor array, a dust concentration sensor, a fan speed measurement device, a power monitoring module and a thermal imager; The temperature sensor array is configured to be evenly distributed at the target position of the heat dissipation area and the sampling frequency of the sensor is set. At the same time, the working condition of the test environment is set to obtain a multidimensional test matrix including M dust concentration levels, N ambient temperature intervals and T heat load levels; Based on the multi-dimensional test matrix, the single fan mode and the multi-fan mode are monitored to obtain temperature change data, dust accumulation data, fan speed data and power consumption data; The temperature change data, the dust accumulation data, the fan speed data and the power consumption data are subjected to outlier detection and filtering to obtain a denoised valid data set, and the denoised valid data set is classified and labeled according to dust concentration, ambient temperature and heat load to obtain a dust environment sample library.

3. The intelligent adjustment method of the intelligent control system of the cooling fan in the dust-resistant environment according to claim 1, characterized in that: The quantitative modeling and analysis of the dust accumulation dynamic characteristics and the air duct blockage formation mechanism based on the dust environment sample library is performed to obtain a thermal equivalent model, including: Performing correlation analysis on the dust accumulation data and operating parameters in the dust environment sample library to obtain a dust accumulation dynamics characteristic model; Gridding the heat dissipation area to obtain a thermal resistance distribution grid, and calculating the thermal resistance of each grid point of the thermal resistance distribution grid to obtain a principle model of air duct blockage formation; Compare and analyze the actual and theoretical cooling efficiencies of cooling fans under different working conditions to obtain a fan performance attenuation model. Perform parameter fitting analysis on the dust accumulation dynamics characteristic model, the air duct blockage formation principle model and the fan performance attenuation model to obtain optimal parameter values ​​including dust adhesion rate coefficient, dust concentration influence coefficient, dust direct influence coefficient and thermal resistance influence coefficient; The optimal parameter values ​​are respectively substituted into the dust accumulation dynamic characteristic model, the air duct blockage formation principle model and the fan performance attenuation model to combine them to obtain a thermal equivalent model.

4. The intelligent adjustment method of the intelligent control system of the cooling fan in an anti-sand and dust environment according to claim 1 is characterized in that: The feature extraction and integration based on the thermal equivalent model to obtain a multi-dimensional dust environment heat dissipation feature vector includes: Screening the instantaneous temperature value, temperature change rate, dust concentration, fan speed and power consumption data output from the thermal equivalent model to obtain an original feature set; The temperature data in the original feature set is processed in segments, and the cumulative sum of the difference between the maximum and minimum temperature in each time window is calculated to obtain a periodic temperature difference sum feature; Continuously recording the power data in the original feature set, calculating a sequence consisting of the standard deviation of the power in each time period, and obtaining an energy consumption fluctuation array feature; Performing Fourier transform analysis on the fan speed data in the original feature set, calculating the ratio of the speed spectrum energy to the original speed signal energy, and obtaining the speed regularity feature; Performing differential and sign change detection on the heat dissipation efficiency data in the original feature set, counting the number of times the heat dissipation efficiency change direction is reversed, and obtaining a heat dissipation efficiency change reversal number feature; The original feature set, the periodic temperature difference sum feature, the energy consumption fluctuation array feature, the rotation speed regularity feature and the heat dissipation efficiency change reversal number feature are combined and dimensionally reduced to obtain a multi-dimensional dust environment heat dissipation feature vector.

5. The intelligent adjustment method of the intelligent control system of the cooling fan in an anti-sand and dust environment according to claim 1, characterized in that: The multi-dimensional dust environment heat dissipation feature vector is input into a gradient boosting learning model with a multi-attraction mechanism for processing to obtain fan collaborative control parameters adapted to the dust environment, including: Inputting the multi-dimensional dust environment heat dissipation feature vector into a deep neural network for feature mapping to obtain mapping feature space data; Inputting the mapped feature space data into a gradient boosting learning model with a multi-attraction mechanism to predict control parameters, and obtaining a preliminary prediction value of the control parameters, wherein the gradient boosting learning model with a multi-attraction mechanism includes a gradient boosting decision module constructed by an XGBoost algorithm; Calculate the similarity between the heat dissipation feature vector of the multi-dimensional dust environment and the predefined K attractors to obtain the similarity value between the feature vector and each attractor; Performing normalization processing based on the similarity values ​​between the feature vector and each attractor to obtain a weight coefficient; The preliminary predicted values ​​of the control parameters corresponding to each attractor are weighted and fused according to the weight coefficients to obtain the fan collaborative control parameters, which include a speed adjustment curve, a power allocation ratio and a response time threshold.

6. The intelligent adjustment method of the intelligent control system of the cooling fan in the dust-resistant environment according to claim 5, characterized in that: The similarity calculation of the multi-dimensional dust environment heat dissipation feature vector and the predefined K attractors is performed to obtain the similarity value between the feature vector and each attractor, including: Performing cluster analysis on historical optimization control strategy data to obtain K typical control strategy center points, wherein the K typical control strategy center points serve as K attractors, and each attractor represents an optimal control mode in the feature space; Normalizing the multi-dimensional dust environment heat dissipation feature vector and the K attractors to obtain a standardized feature vector and a standardized attractor; Calculating the Euclidean distance between the standardized feature vector and each standardized attractor to obtain K distance values, and performing Gaussian kernel function transformation on each distance value to obtain an initial similarity value; The initial similarity value is dynamically adjusted based on the dust concentration and the load level to obtain the similarity value between the feature vector and each attractor.

7. The intelligent adjustment method of the intelligent control system of the cooling fan in an anti-sand and dust environment according to claim 1, characterized in that: The intelligent adjustment method of the intelligent control system of the heat dissipation fan in the anti-sand and dust environment also includes: Defining the phase of each fan in the multi-fan system to obtain a fan control phase parameter, wherein the fan control phase parameter satisfies that the phase difference between adjacent fans is 2π divided by the total number of fans; Calculating the basic speed of each fan based on the speed adjustment curve in the fan cooperative control parameter to obtain a speed reference value, and performing sinusoidal modulation on the speed of each fan according to the fan control phase parameter to make each fan form an alternating working state; Dynamically adjusting the power allocation ratio in the fan cooperative control parameter according to the temperature difference of the areas responsible for each fan to obtain a temperature-adaptive power allocation ratio; The dust adaptive response time is obtained by calculating the dust accumulation amount of each fan based on the response time threshold in the fan collaborative control parameter; The rate of change of dust concentration in the environment is monitored in real time. When a sudden change in dust concentration is detected, the emergency adjustment mechanism is triggered to obtain a speed curve optimized for dust prevention. The temperature-adaptive power allocation ratio, the dust-adaptive response time and the dust-optimized speed curve are combined and applied to the control of each fan to obtain a complementary fan control strategy.

8. An intelligent control system intelligent adjustment device for a cooling fan in an anti-dust environment, characterized in that: The method for intelligently adjusting the cooling fan intelligent control system in an anti-sand and dust environment according to any one of claims 1 to 7, wherein the intelligent adjusting device for the cooling fan intelligent control system in an anti-sand and dust environment comprises: The acquisition module is used to collect data of multiple working conditions on the heat dissipation test platform of the heat dissipation fan to obtain a dust environment sample library; A modeling module, used to perform quantitative modeling and analysis on the dust accumulation dynamics characteristics and the air duct blockage formation mechanism based on the dust environment sample library to obtain a thermal equivalent model; A feature extraction module, used for extracting and integrating features based on the thermal equivalent model to obtain a multi-dimensional dust environment heat dissipation feature vector; The processing module is used to input the multi-dimensional dust environment heat dissipation feature vector into a gradient boosting learning model with a multi-attraction mechanism for processing, so as to obtain fan cooperative control parameters adapted to the dust environment.

Images