A mining lamp brightness adjustment method based on support vector regression model

Abstract

The present invention discloses a mining lamp brightness adjustment method based on a support vector regression model. The mining lamp brightness adjustment method of the present invention can move the light source and automatically adjust to the optimal light source brightness that best matches the current use environment in a complex mine environment. This automatic adjustment can not only perfectly meet the lighting needs during mine operations and ensure the visual clarity and operational safety of operators, but also minimize energy waste, effectively extend the power supply time of sodium ion batteries, and improve discharge efficiency, thereby achieving the longest working time, providing stable, reliable and energy-saving lighting support for mine operations, greatly improving the efficiency and safety of mine operations, and reducing energy costs and equipment maintenance costs. This technology has broad application prospects and significant economic benefits in the field of mine lighting.

Description

Technical Field

[0001] The invention belongs to the field of brightness adjustment of mining lamps, and in particular, relates to a method for adjusting the brightness of mining lamps based on a support vector regression model. Background Art

[0002] With the continuous strengthening of mine safety management, the design and function of mining lamps, as a key tool for miners' safe operation, are developing towards intelligence and efficiency. However, the existing mining lamps still have some significant problems in the complex working environment of mines, which are as follows:

[0003] Limited adjustable gears and insufficient brightness options: Traditional mining lamps have few adjustable gears and provide very limited brightness options, which cannot flexibly adapt to the diverse operating scene requirements in mines. For example, in narrow equipment maintenance areas, miners need high-brightness diffuse lighting, while when inspecting wide coal mining surfaces, strong light needs to be focused to illuminate the distant working environment. Due to the limitations of gear selection, it is difficult for mining lamps to accurately meet the lighting needs in these scenarios, which not only affects miners' clear observation of the surrounding environment and operating objects, reducing work efficiency, but may also cause operational errors due to insufficient or inappropriate lighting, thereby creating safety hazards.

[0004] Unable to control light intensity in real time: The mine environment is complex and changeable. As miners move and work progress, the lighting requirements will continue to change. However, traditional mining lamps cannot achieve real-time and precise control of light intensity, and miners can only passively use lighting with fixed brightness. This limitation of fixed light intensity may lead to a series of problems, such as glare caused by excessive light that affects vision, or inability to clearly see the working environment in the distance or in the dark due to too weak light, thus affecting the smooth progress of mining operations and may even pose safety hazards.

[0005] Energy waste and battery life issues: Since mining lamps cannot flexibly adjust the light intensity according to actual needs, miners often have to use a higher-level fixed brightness when only a lower brightness is needed to meet the work needs, which undoubtedly leads to unnecessary energy waste. Energy waste accelerates the consumption of mining lamp batteries and shortens their battery life. It may cause the mining lamp to run out of power before the end of the work shift, causing inconvenience to miners and affecting the orderly connection of mine work.

[0006] In view of this, the present invention is proposed. Summary of the invention

[0007] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a mining lamp brightness adjustment method based on a support vector regression model, thereby solving the problems raised in the above-mentioned background technology.

[0008] In order to solve the above technical problems, the basic concept of the technical solution adopted by the present invention is:

[0009] A method for adjusting the brightness of a mining lamp based on a support vector regression model comprises the following steps:

[0010] S1: Real-time monitoring of environmental parameters in the mine through multiple environmental sensors, including air pressure, humidity, light intensity, UV intensity and dust concentration;

[0011] S2: transmitting the real-time data collected by the environmental sensor to the data processing unit, and ensuring the stability and timeliness of the data;

[0012] S3: Preprocess the received data, including data cleaning, normalization, and feature construction, to remove outliers and ensure data consistency;

[0013] S4: Processing the preprocessed data using a regression model to generate a regression function for adjusting the brightness of the mining lamp and optimizing the battery energy output according to environmental parameters;

[0014] S5: receiving the regression function control instruction in step S4, and automatically adjusting the brightness level of the mining lamp through the control circuit of the mining lamp, and adjusting the light source intensity according to environmental changes to meet the needs of different working areas.

[0015] Optionally, the multiple environmental sensors include an air pressure sensor, a humidity sensor, a photoresistor sensor, an ultraviolet sensor and a laser particle sensor, which are used to collect environmental data of various areas in the mine in real time.

[0016] Optionally, the step of preprocessing the received data is

[0017] Check all kinds of collected raw data and remove abnormal data points, including environmental data that does not conform to a reasonable range due to factors such as sensor failure or external interference;

[0018] The processed environmental data is normalized, which converts environmental parameters of different dimensions and magnitudes into a unified scale for subsequent model training. Normalization includes but is not limited to the conversion of data such as air pressure, humidity, light intensity, ultraviolet intensity, and dust concentration, so that the influence weights of various environmental data can be evenly distributed to avoid deviations in model training effects due to magnitude differences.

[0019] Analyze the potential correlation between various environmental data parameters and the brightness adjustment of mining lamps and battery energy output to screen out key features. According to the correlation between environmental data and target variables, remove redundant features and extract features that contribute greatly to model performance. Then, construct new features, which are generated by combining and calculating existing environmental parameters.

[0020] Optionally, the regression model is trained before processing the preprocessed data using the regression model, the steps are:

[0021] Selecting a radial basis kernel function for a support vector regression model, determining key parameters that affect the model training effect based on the selected radial basis kernel function, including but not limited to a width parameter and a penalty parameter of the kernel function, and optimizing them through a cross-validation method;

[0022] The preprocessed data set is divided into training set, validation set and test set. The model is initially trained on the training set, and the model parameters are further optimized through error analysis on the validation set.

[0023] Input the data in the training set into the support vector regression model, use the iterative algorithm to train according to the selected kernel function and optimized parameters, adjust the weight coefficients in the model, and fit the relationship between the input features and the target variable;

[0024] A support vector regression model is defined according to a selected kernel function, wherein the kernel function transforms the data from a low-dimensional space to a high-dimensional space by mapping the input data, so that a linear regression method can be used in the high-dimensional space to fit the data relationship;

[0025] Through the trained SVR model, a regression function is generated according to the input environmental data, which is used to control the brightness of the mining lamp and optimize the battery energy output.

[0026] Optionally, after the model training is completed, the model needs to be evaluated. The steps are as follows:

[0027] Select and error, mean absolute error, and coefficient of determination to measure the performance of the trained SVR model. Then, based on the calculated evaluation indicators, determine whether the performance of the model meets expectations. If the evaluation results show that the accuracy of the model is insufficient, enter the model optimization adjustment step;

[0028] When optimizing and adjusting the model according to the evaluation results, the optimization and adjustment include but are not limited to: re-examining the data preprocessing process, adjusting the feature selection and construction method, optimizing the model parameters, and during the optimization process, using appropriate cross-validation to retrain the optimized model to reduce the prediction error of the model on the validation set;

[0029] By repeating the optimization and evaluation process, an SVR model with stable performance and high accuracy is finally obtained.

[0030] Optionally, the mining lamp is also integrated with an ARM architecture control board that supports a vector regression model. The control board is used to control the lighting level and battery output power of the mining lamp device according to the monitoring results of the external environment, and to adjust the light source brightness and battery energy output through two control methods.

[0031] Optionally, after the model evaluation structure is correct, the model needs to be stored in the ARM control board of the mining lamp. The steps are:

[0032] Extract key parameters from the trained support vector machine model, including support vectors, corresponding coefficients and biases, and organize these parameters into a structure array format suitable for the ARM platform;

[0033] According to the hardware characteristics of the ARM platform, the extracted floating-point model parameters are converted into fixed-point representation. Then, the SVM model operation code originally written in Python is converted into C++ code suitable for the ARM platform. The calculation speed is improved by optimizing vector calculations and using ARM's NEON instruction set.

[0034] Design and implement a data interface suitable for the ARM platform to receive external data input and transmit it to the SVM model for processing. At the same time, ensure that the model output results can be transmitted to the mining lamp equipment through the communication interface;

[0035] The sorted SVM model parameters are loaded into the external memory of the ARM platform, including static random access memory and dynamic random access memory, and the related variables and pointers are initialized.

[0036] Optionally, the steps of receiving the regression function control instruction in step S4, automatically adjusting the brightness level of the mining lamp through the control circuit of the mining lamp, and adjusting the light source intensity according to environmental changes to adapt to the needs of different working areas are:

[0037] Based on the time-ordered environmental parameter data set, the data is transmitted to the ARM architecture control board with integrated vector regression model for analyzing and processing environmental data to generate decisions for lighting adjustment and battery optimization;

[0038] According to the analysis results of the ARM control board, the mining lamp equipment automatically switches the lighting level and adjusts the brightness. Specifically, in areas with darker light, the control board instructs the mining lamp to switch to a higher lighting level and increase the brightness; in areas with more sufficient light, it automatically switches to a lower lighting level and reduces the brightness.

[0039] While adjusting the light brightness, the ARM control board optimizes the discharge strategy of the mining lamp battery according to the current ambient light conditions and brightness adjustment to reduce the waste of battery energy.

[0040] According to the current ambient brightness of the mining lamp, the ARM control board intelligently adjusts the battery output power. In areas with moderate light, the control board reduces the battery output power to ensure that the battery can work at a lower and more stable power output, thereby maximizing the battery life and ensuring that the mining lamp can continue to provide sufficient lighting;

[0041] When the mining lamp equipment completes the task, the ARM control board further optimizes the battery usage according to the battery status and actual needs, prolongs the battery life, and ensures stable lighting services throughout the work shift until the task is completed.

[0042] After adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art. Of course, any product implementing the present invention does not necessarily need to achieve all the advantages described below at the same time:

[0043] The mining lamp brightness adjustment method of the present invention can move the light source and automatically adjust to the optimal light source brightness that best matches the current use environment in a complex mine environment. This automatic adjustment can not only perfectly meet the lighting needs during mine operations and ensure the visual clarity and operational safety of operators, but also minimize energy waste, effectively extend the power supply time of sodium ion batteries, and improve discharge efficiency, thereby achieving the longest working time, providing stable, reliable and energy-saving lighting support for mine operations, greatly improving the efficiency and safety of mine operations, and reducing energy costs and equipment maintenance costs. This technology has broad application prospects and significant economic benefits in the field of mine lighting.

[0044] The specific implementation modes of the present invention are further described in detail below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The drawings described below are only some embodiments. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0046] In the figure:

[0047] Figure 1 This is a schematic diagram of the operation of the mining lamp;

[0048] Figure 2 This is a schematic diagram of data collection;

[0049] Figure 3 It is a schematic diagram of the data processing flow;

[0050] Figure 4 This is a schematic diagram of the mining lamp control method;

[0051] Figure 5 Schematic diagram of the algorithm control process.

[0052] It should be noted that these drawings and textual descriptions are not intended to limit the conceptual scope of the present invention in any way, but are intended to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0053] The present invention will now be described in further detail with reference to the accompanying drawings.

[0054] See also Figure 1-5 As shown, in this embodiment, a method for adjusting the brightness of a mining lamp based on a support vector regression model is provided, comprising the following steps:

[0055] S1: Real-time monitoring of environmental parameters in the mine through multiple environmental sensors, including air pressure, humidity, light intensity, UV intensity and dust concentration;

[0056] S2: transmitting the real-time data collected by the environmental sensor to the data processing unit, and ensuring the stability and timeliness of the data;

[0057] S3: Preprocess the received data, including data cleaning, normalization, and feature construction, to remove outliers and ensure data consistency;

[0058] S4: Use the regression model to process the preprocessed data and generate a regression function for adjusting the brightness of the mining lamp and optimizing the battery energy output according to environmental parameters; the generated regression function adjusts the light source brightness of the mining lamp to ensure that it changes according to environmental conditions and optimizes the output power of the battery according to the regression function to ensure efficient use of battery energy and extend the working time of the mining lamp.

[0059] S5: receiving the regression function control instruction in step S4, and automatically adjusting the brightness level of the mining lamp through the control circuit of the mining lamp, and adjusting the light source intensity according to environmental changes to meet the needs of different working areas.

[0060] like Figure 2 As shown, in this embodiment, the multiple environmental sensors include an air pressure sensor, a humidity sensor, a photoresistor sensor, an ultraviolet sensor and a laser particle sensor, which are used to collect environmental data of various areas in the mine in real time.

[0061] Air pressure data collection

[0062] The air pressure in different areas of the mine is monitored by analog air pressure sensors. The monitored air pressure range is up to 10-115kPa, the accuracy reaches 0.1%, and it can work stably at an operating temperature of -40-85℃.

[0063] At the mine entrance, the air pressure is close to the local atmospheric pressure, usually about 101.325kPa; in the main air intake tunnel, the air pressure range is generally 102kPa-108kPa; the air pressure in the working area fluctuates around 100kPa-106kPa; the air pressure in the return air tunnel is relatively low, ranging from 95kPa to 100kPa. The data from these different areas will be continuously collected at a certain time interval to form a time series data set about air pressure, and because the working area needs to monitor small fluctuations in air pressure, the high sensitivity of the sensor ensures the accuracy of the data here.

[0064] Humidity data collection

[0065] The humidity sensor is used to detect the ambient humidity. The relative humidity monitoring range covers 0-100%, the accuracy is ±2%, and the operating temperature range is -40-85℃.

[0066] Starting from the mine entrance, which is basically the same as the outside atmospheric humidity, different areas of the underground mine present different humidity conditions. For example, the relative humidity of the shallow mine tunnel with good ventilation is about 60%-70%; the relative humidity of the air intake tunnel is about 60%-80%; the relative humidity of the deep mine tunnel can reach 80%-95%; the relative humidity of the mine working face is in a saturated state, greater than 95%. The humidity sensor also continuously collects humidity data in each area at a high frequency to provide a comprehensive humidity information sample for subsequent analysis.

[0067] Lighting data collection

[0068] A photoresistor sensor is used to detect ambient light. It has a withstand voltage of 150V, a maximum power of 90MW, an adaptable ambient temperature range of -30-70℃, a central wavelength of 540NM, a bright resistance of 5-10kΩ, and a dark resistance of 0.2Ω.

[0069] In the lighting environment of explosion-proof LED mining lamps (whose spectral wavelength is between 450NM-465NM, and the resistance value of the photoresistor changes in a small and constant range under its illumination), and in combination with different lighting conditions in the mine, the photoresistor sensor continuously obtains data related to light intensity to distinguish between different lighting conditions in different scenarios, such as sunlight on the ground, unlit nights, and underground mines illuminated only by explosion-proof mining lamps.

[0070] UV data collection

[0071] Use photodiode to detect ultraviolet intensity, response wavelength 200 ~ 370NM, accuracy ± 1UV INDEX, working temperature -20 ~ 85 ℃, annual drift rate <5%.

[0072] It has the advantages of high sensitivity and wide wavelength response range. It can convert the incident ultraviolet energy into a current signal, and process and measure it through an amplifying circuit. Use ultraviolet sensors to detect the ultraviolet conditions in the corresponding environment. By judging whether the location is on the ground with sunlight, at night without lighting, or in an underground mine illuminated only by explosion-proof mining lamps, etc., collect data on ultraviolet intensity in different scenes, and then clarify the changing characteristics of ultraviolet rays in different environments, providing a basis for overall environmental assessment.

[0073] Dust data collection

[0074] The dust sensor is used to monitor the dust situation in the mine. It can detect the minimum dust particle diameter in the range of 0.3-10um and the particle concentration range is 0-1000ug / m 3 The particle mass concentration resolution reaches 1ug / m 3 , the operating temperature range is -10-60℃.

[0075] According to the characteristics of dust particles of different particle sizes in the mine, such as the respective settling characteristics of coarse dust, fine dust, micro dust, and ultra-fine dust, as well as the national standard for the respirable dust concentration of different types of dust after dust settling in the mine (coal dust, rock dust, etc. are distinguished based on the free silicon dioxide content), the dust sensor collects relevant data such as dust concentration in each area in an all-round manner to reflect the actual situation of the dust environment in the mine.

[0076] The diameter of dust particles in the mine air is coarse dust: the particle size is greater than 40μm, these particles are easy to settle in the air.

[0077] Fine dust: The particle size is between 10 and 40 μm, visible to the naked eye, and will settle faster in still air.

[0078] Fine dust: The particle size is between 0.25 and 10 μm, needs to be observed through an optical microscope, and settles at a constant speed in still air.

[0079] Ultrafine dust: particle size is less than 0.25μm, needs to be observed by electron microscope, diffusion dynamics

[0080] According to the national standard for dust concentration after mine dust reduction, for coal dust (free silica content less than 10%), the respirable dust concentration shall not exceed 2.5 mg / m 3 ;

[0081] For rock dust, when the free silica content is between 5% and 10%, the respirable dust concentration shall not exceed 1 mg / m 3 When the free silica content is between 10% and 30%, the respirable dust concentration shall not exceed 0.5 mg / m 3 When the free silica content is between 30% and 50%, the respirable dust concentration shall not exceed 0.3 mg / m3 When the free silica content is greater than 50%, the respirable dust concentration shall not exceed 0.2 mg / m 3 .

[0082] like Figure 3 As shown, in this embodiment, the steps of preprocessing the received data are:

[0083] Check all types of collected raw data and remove abnormal data points, including environmental data that does not conform to the reasonable range due to sensor failure or external interference, such as air pressure, humidity, light and other data. For example, if the humidity value collected by the humidity sensor at a certain moment exceeds its set range, or the light intensity data obviously deviates from the normal fluctuation range, then identify and remove these data to ensure data quality.

[0084] The processed environmental data is normalized, which converts environmental parameters of different dimensions and magnitudes into a unified scale for subsequent model training. Normalization includes but is not limited to the conversion of data such as air pressure, humidity, light intensity, ultraviolet intensity, and dust concentration, so that the influence weights of various environmental data can be evenly distributed to avoid deviations in model training effects due to magnitude differences.

[0085] Analyze the potential correlation between various environmental data parameters and the brightness adjustment of mining lamps and battery energy output, so as to screen out key features. According to the correlation between environmental data and target variables, remove redundant features, and extract features that greatly contribute to model performance. For example, the interaction between air pressure and light intensity has a greater impact on the brightness demand of light sources, and the combination of humidity and dust concentration has an important impact on the optimization of battery discharge strategy. Next, construct new features, which are generated by combining and calculating existing environmental parameters. For example, calculate the ratio of air pressure to humidity to reflect the comprehensive state of the air, or combine light intensity and ultraviolet intensity to construct features that represent the quality of ambient light. The feature construction further enhances the expressive power of the data set, enabling subsequent models to better understand and predict the needs of mining lamp brightness adjustment and battery energy output.

[0086] like Figure 3 As shown, in this embodiment, the regression model is trained before the pre-processed data is processed using the regression model, and the steps are as follows:

[0087] Selecting a radial basis kernel function (RBF) for a support vector regression (SVR) model, determining key parameters that affect the model training effect based on the selected radial basis kernel function (RBF), including but not limited to the kernel function width parameter (gamma value) and penalty parameter (C value), and optimizing them through a cross-validation method;

[0088] The preprocessed data set is divided into training set, validation set and test set. The model is initially trained on the training set, and the model parameters are further optimized through error analysis on the validation set to ensure good generalization ability of the model and avoid overfitting or underfitting.

[0089] The data in the training set (including the screened and constructed environmental features and the corresponding target variables, such as the optimal light source brightness and battery energy output) are input into the support vector regression (SVR) model. According to the selected kernel function and optimized parameters, the iterative algorithm is used for training, the weight coefficients in the model are adjusted, and the relationship between the input features and the target variables is fitted.

[0090] A support vector regression (SVR) model is defined according to the selected kernel function, wherein the kernel function transforms the data from a low-dimensional space to a high-dimensional space by mapping the input data, so that a linear regression method can be used in the high-dimensional space to fit the data relationship;

[0091] Establish a loss function, wherein the loss function includes an ε-insensitive loss, which allows an error less than or equal to ε between the predicted value and the true value without incurring a loss, and the loss will only increase when the error is greater than ε;

[0092] The SVR model is optimized by minimizing the objective function, which consists of two parts: one is the regularization term, which is used to control the complexity of the model; the other is the error term, which is used to minimize the error of the model on the training set. The objective function is as follows:

[0093] Among them, |w| 2 is the square of the weight vector, representing the model complexity; ξ i is a slack variable used to measure the error of each data point; C is a regularization parameter used to balance error and complexity.

[0094] Apply the kernel function to map the input data to a high-dimensional space. The kernel function can be a radial basis function (RBF) kernel, and its expression is: K(x i , x j )=exp(-γ|x i -x j | 2 ), where γ is the kernel function parameter, which controls the similarity between data points. The kernel function is used to process the nonlinear relationship between the input features and the target variable;

[0095] Through the trained SVR model, a regression function is generated based on the input environmental data (including air pressure, humidity, light intensity, ultraviolet intensity, dust concentration, etc.). The regression function is used to control the brightness of the mining lamp and optimize the battery energy output, thereby ensuring the lighting effect and energy utilization efficiency of the mining lamp are maximized. The expression of the regression function is: Where f(x) is the predicted value of the model, which means the output predicted by the model (i.e., brightness adjustment value or battery output power) when the input data x is given, x i is the support vector data point in the training set, and α is the support vector x i The Lagrange multiplier of represents the contribution of the support vector to the final prediction, K(x, x i ) is the kernel function, which uses the radial basis function (RBF) kernel to represent the input x and the training data point x i The similarity between them is the bias term, which represents the intercept of the regression function.

[0096] In this embodiment, after the model training is completed, the model needs to be evaluated, and the steps are as follows:

[0097] The error, mean absolute error, and determination coefficient are selected to measure the performance of the trained SVR model. The evaluation indicators include mean square error (MSE), mean absolute error (MAE), determination coefficient (R2), etc., which are used to comprehensively evaluate the model's prediction accuracy, error range, and its ability to explain changes in the target variable. Next, based on the calculated evaluation indicators, determine whether the performance of the model meets expectations. If the evaluation results show that the accuracy of the model is insufficient (for example, the mean square error is too large or the determination coefficient is low), enter the model optimization adjustment step.

[0098] When optimizing and adjusting the model according to the evaluation results, the optimization and adjustment include but are not limited to: re-examining the data preprocessing process, adjusting feature selection and construction methods, and optimizing model parameters. Through these adjustments, the performance of the model is continuously improved to ensure that it can accurately reflect the relationship between changes in the mine environment and the brightness of the light source and the energy output of the battery. During the optimization process, appropriate cross-validation is used to retrain the optimized model to reduce the prediction error of the model on the validation set, and to have good generalization ability to avoid overfitting or underfitting.

[0099] Through repeated optimization and evaluation processes, we finally obtained an SVR model with stable performance and high accuracy. This model can accurately generate a control regression function based on environmental data such as air pressure, humidity, light intensity, ultraviolet intensity, and dust concentration, thereby achieving precise control of the brightness of the mining lamp and effective optimization of battery energy output.

[0100] like Figure 3As shown, in this embodiment, an ARM architecture control board of a support vector regression (SVR) model is also integrated in the mining lamp. The control board is used to control the lighting level and battery output power of the mining lamp device according to the monitoring results of the external environment, and to adjust the light source brightness and battery energy output through two control methods.

[0101] The control board can adjust the lighting gear of the mining lamp equipment through a manual control switch, and the manual control switch includes two gears: focused strong light and scattered light. Focused strong light is suitable for scenes that require concentrated light, while scattered light is suitable for scenes that require scattered light. In addition, the algorithm can also control the switch to monitor external environmental data, and the external environmental data includes parameters such as air pressure, humidity, light intensity, ultraviolet intensity, and particle concentration. The ARM architecture control board adjusts the lighting gear of the mining lamp and the battery output power according to the changes in sensor data. Subsequently, according to the environmental data collected from various sensors, the ARM architecture control board adjusts the lighting gear of the mining lamp equipment and the output power of the battery management system (BMS) through the gear switching circuit and the power output circuit to ensure that the lighting intensity of the mining lamp adapts to the specific working environment of the mine. Finally, according to changes in the external environment, the ARM architecture control board can switch the lighting mode of the mining lamp according to data input, including adjusting the lighting intensity of focused strong light or adjusting the lighting intensity of scattered light, to ensure that the mining lamp equipment can provide appropriate light source intensity according to different working conditions and needs.

[0102] In this embodiment, after the model evaluation structure is correct, the model needs to be stored in the ARM control board of the mining lamp, and the steps are as follows:

[0103] Extract key parameters from the trained support vector machine (SVM) model, including support vectors, corresponding coefficients, and biases, and organize these parameters into a structure array format suitable for the ARM platform to ensure that they can be efficiently called in subsequent operations;

[0104] According to the hardware characteristics of the ARM platform, the extracted floating-point model parameters (coefficients of support vectors) are converted into fixed-point representation to improve the operation efficiency and ensure the best performance in scenarios with strict resource and power requirements. Then, the SVM model operation code originally written in Python is converted into C++ code suitable for the ARM platform. The calculation speed is improved by optimizing the vector calculation and using the ARM NEON instruction set.

[0105] Design and implement a data interface suitable for the ARM platform to receive external data input and transmit it to the SVM model for processing. At the same time, ensure that the model output results can be transmitted to the mining lamp equipment through communication interfaces such as UART, SPI, USB, etc.

[0106] The sorted SVM model parameters are loaded into the external memory of the ARM platform, including static random access memory (SRAM) and dynamic random access memory (DRAM), and the relevant variables and pointers are initialized to ensure that these parameters can be accurately called and calculated in subsequent operations.

[0107] like Figure 4-5 As shown, in this embodiment, the steps of receiving the regression function control instruction in step S4, automatically adjusting the brightness level of the miner's lamp through the control circuit of the miner's lamp, and adjusting the light source intensity according to environmental changes to adapt to the needs of different working areas are:

[0108] Based on the time-ordered environmental parameter data set, the data is transmitted to the ARM architecture control board with integrated vector regression model for analyzing and processing environmental data to generate decisions for lighting adjustment and battery optimization;

[0109] According to the analysis results of the ARM control board, the mining lamp equipment automatically switches the lighting level and adjusts the brightness. Specifically, in areas with darker light, the control board instructs the mining lamp to switch to a higher lighting level and increase the brightness; in areas with more sufficient light, it automatically switches to a lower lighting level and reduces the brightness.

[0110] While adjusting the light brightness, the ARM control board optimizes the discharge strategy of the mining lamp battery according to the current ambient light conditions and brightness adjustment to reduce the waste of battery energy.

[0111] According to the current ambient brightness of the mining lamp, the ARM control board intelligently adjusts the battery output power. In areas with moderate light, the control board reduces the battery output power to ensure that the battery can work at a lower and more stable power output, thereby maximizing the battery life and ensuring that the mining lamp can continue to provide sufficient lighting;

[0112] When the mining lamp equipment completes the task, the ARM control board further optimizes the battery usage according to the battery status and actual needs, prolongs the battery life, and ensures stable lighting services throughout the work shift until the task is completed.

[0113] For example:

[0114] Assuming that the miner's lamp is used for mine operations, miners need to work in the mine for about 8 hours. During this period, the miner's lamp needs to provide sufficient light, and the battery needs to avoid excessive consumption while ensuring the lighting effect and extending the use time. To achieve this goal, the ARM control board will perform the following optimizations:

[0115] In the early days of the operation of the mining lamp equipment, when miners enter the darker deep mine area, the external light is insufficient, and the mining lamp needs to provide higher brightness to ensure safe operation. At this time, the ARM control board will automatically adjust the mining lamp to a higher light level based on the real-time collected environmental data, such as light intensity and mine depth, to ensure that miners can work normally in a dim environment. Although higher brightness is required, the ARM control board will also optimize the battery discharge strategy to avoid excessive battery consumption in a short period of time and maintain sufficient power to meet subsequent needs.

[0116] As miners move to areas with better lighting conditions in the mine, such as the air intake tunnel, the ambient light gradually becomes sufficient, and the brightness requirement of the mining lamp decreases. The ARM control board will automatically adjust the mining lamp to a lower lighting level according to changes in environmental data to reduce unnecessary energy waste. At the same time, the output power of the battery will also be intelligently adjusted to a lower level to maintain a low and stable power output, thereby extending the battery life and ensuring that the mining lamp continues to work stably.

[0117] When the mining lamp is nearing the end of its operation, the ambient light becomes dimmer and the remaining battery power gradually decreases. At this time, the ARM control board will intelligently adjust the light intensity according to the real-time battery status and environmental changes to ensure that the mining lamp can provide the required lighting and avoid excessive battery discharge. For example, in areas with high light demand, the ARM control board will try to extend the remaining battery life by reducing the brightness or adjusting the battery output strategy to ensure that the mining lamp can continue to provide sufficient light until the task is completed.

[0118] After the mining lamp completes the entire operation task, the ARM control board will optimize the battery management to ensure that the battery is not over-depleted. At this time, the battery output power will be adjusted to the minimum to protect the battery and ensure its long-term stable operation.

[0119] The present invention is not limited to the above-mentioned embodiments. Anyone should be aware that any structural changes made under the enlightenment of the present invention, and any technical solutions that are the same or similar to the present invention, fall within the protection scope of the present invention. The technology, shape, and structural parts not described in detail in the present invention are all well-known technologies.

Claims

1. A method for adjusting the brightness of a mining lamp based on a support vector regression model, characterized in that: The following steps are involved: S1: Real-time monitoring of environmental parameters in the mine through multiple environmental sensors, including air pressure, humidity, light intensity, UV intensity and dust concentration; S2: transmitting the real-time data collected by the environmental sensor to the data processing unit, and ensuring the stability and timeliness of the data; S3: Preprocess the received data, including data cleaning, normalization, and feature construction, to remove outliers and ensure data consistency; S4: Processing the preprocessed data using a regression model to generate a regression function for adjusting the brightness of the mining lamp and optimizing the battery energy output according to environmental parameters; S5: receiving the regression function control instruction in step S4, and automatically adjusting the brightness level of the mining lamp through the control circuit of the mining lamp, and adjusting the light source intensity according to environmental changes to meet the needs of different working areas.

2. A method for adjusting the brightness of a mining lamp based on a support vector regression model according to claim 1, characterized in that: The multiple environmental sensors include an air pressure sensor, a humidity sensor, a photoresistor sensor, an ultraviolet sensor and a laser particle sensor, which are used to collect environmental data of various areas in the mine in real time.

3. The method for adjusting the brightness of a mining lamp based on a support vector regression model according to claim 1, characterized in that: The steps for preprocessing the received data are: Check all kinds of collected raw data and remove abnormal data points, including environmental data that does not conform to a reasonable range due to factors such as sensor failure or external interference; The processed environmental data is normalized, which converts environmental parameters of different dimensions and magnitudes into a unified scale for subsequent model training. Normalization includes but is not limited to the conversion of data such as air pressure, humidity, light intensity, ultraviolet intensity, and dust concentration, so that the influence weights of various environmental data can be evenly distributed to avoid deviations in model training effects due to magnitude differences. Analyze the potential correlation between various environmental data parameters and the brightness adjustment of mining lamps and battery energy output to screen out key features. According to the correlation between environmental data and target variables, remove redundant features and extract features that contribute greatly to model performance. Then, construct new features, which are generated by combining and calculating existing environmental parameters.

4. The method for adjusting the brightness of a mining lamp based on a support vector regression model according to claim 1, characterized in that: Before using the regression model to process the preprocessed data, the regression model is trained. The steps are as follows: Selecting a radial basis kernel function for a support vector regression model, determining key parameters that affect the model training effect based on the selected radial basis kernel function, including but not limited to a width parameter and a penalty parameter of the kernel function, and optimizing them through a cross-validation method; The preprocessed data set is divided into training set, validation set and test set. The model is initially trained on the training set, and the model parameters are further optimized through error analysis on the validation set. Input the data in the training set into the support vector regression model, use the iterative algorithm to train according to the selected kernel function and optimized parameters, adjust the weight coefficients in the model, and fit the relationship between the input features and the target variable; A support vector regression model is defined according to a selected kernel function, wherein the kernel function transforms the data from a low-dimensional space to a high-dimensional space by mapping the input data, so that a linear regression method can be used in the high-dimensional space to fit the data relationship; Through the trained SVR model, a regression function is generated according to the input environmental data, which is used to control the brightness of the mining lamp and optimize the battery energy output.

5. The method for adjusting the brightness of a mining lamp based on a support vector regression model according to claim 1, characterized in that: After completing the model training, the model needs to be evaluated. The steps are: Select and error, mean absolute error, and determination coefficient to measure the performance of the trained SVR model. Then, based on the calculated evaluation indicators, determine whether the performance of the model meets expectations. If the evaluation results show that the accuracy of the model is insufficient, enter the model optimization adjustment step; When optimizing and adjusting the model according to the evaluation results, the optimization and adjustment include but are not limited to: re-examining the data preprocessing process, adjusting the feature selection and construction method, optimizing the model parameters, and during the optimization process, using appropriate cross-validation to retrain the optimized model to reduce the prediction error of the model on the validation set; By repeating the optimization and evaluation process, an SVR model with stable performance and high accuracy is finally obtained.

6. The method for adjusting the brightness of a mining lamp based on a support vector regression model according to claim 1, characterized in that: The mining lamp also integrates an ARM architecture control board that supports a vector regression model. The control board is used to control the lighting level and battery output power of the mining lamp equipment according to the monitoring results of the external environment, and to adjust the light source brightness and battery energy output through two control methods.

7. The method for adjusting the brightness of a mining lamp based on a support vector regression model according to claim 1, characterized in that: After the model evaluation structure is correct, the model needs to be stored in the ARM control board of the mining lamp. The steps are as follows: Extract key parameters from the trained support vector machine model, including support vectors, corresponding coefficients and biases, and organize these parameters into a structure array format suitable for the ARM platform; According to the hardware characteristics of the ARM platform, the extracted floating-point model parameters are converted into fixed-point representation. Then, the SVM model operation code originally written in Python is converted into C++ code suitable for the ARM platform. The calculation speed is improved by optimizing vector calculations and using ARM's NEON instruction set. Design and implement a data interface suitable for the ARM platform to receive external data input and transmit it to the SVM model for processing. At the same time, ensure that the model output results can be transmitted to the mining lamp equipment through the communication interface; The sorted SVM model parameters are loaded into the external memory of the ARM platform, including static random access memory and dynamic random access memory, and the related variables and pointers are initialized.

8. The method for adjusting the brightness of a mining lamp based on a support vector regression model according to claim 1, characterized in that: The steps of receiving the regression function control instruction in step S4, automatically adjusting the brightness level of the mining lamp through the control circuit of the mining lamp, and adjusting the light source intensity according to environmental changes to adapt to the needs of different working areas are as follows: Based on the time-ordered environmental parameter data set, the data is transmitted to the ARM architecture control board with integrated vector regression model for analyzing and processing environmental data to generate decisions for lighting adjustment and battery optimization; According to the analysis results of the ARM control board, the mining lamp equipment automatically switches the lighting level and adjusts the brightness. Specifically, in dark areas, the control board instructs the mining lamp to switch to a higher lighting level and increase the brightness; In areas with more light, it automatically switches to a lower light level and reduces brightness. While adjusting the light brightness, the ARM control board optimizes the discharge strategy of the mining lamp battery according to the current ambient light conditions and brightness adjustment to reduce the waste of battery energy. According to the current ambient brightness of the mining lamp, the ARM control board intelligently adjusts the battery output power. In areas with moderate light, the control board reduces the battery output power to ensure that the battery can work at a lower and more stable power output, thereby maximizing the battery life and ensuring that the mining lamp can continue to provide sufficient lighting; When the mining lamp equipment completes the task, the ARM control board further optimizes the battery usage according to the battery status and actual needs, prolongs the battery life, and ensures stable lighting services throughout the work shift until the task is completed.

Images