Inhibition analysis method for blurred screen of liquid crystal display screen

Abstract

The invention provides a blurred screen suppression analysis method for a liquid crystal display screen, and the method comprises the steps: processing the real-time working state information of the liquid crystal display screen through employing a support vector machine algorithm, predicting the power demand of a drive circuit, and obtaining a power demand curve of the drive circuit in different working states; whether a blurred screen phenomenon occurs or not is judged according to the real-time working state of the liquid crystal display screen, when the blurred screen is detected, the working intensity and the power budget of the suppression circuit are dynamically improved to eliminate the blurred screen, and when display is normal, the suppression intensity is reduced; the use environment parameters of the liquid crystal display screen are acquired in real time, the power distribution modes of the driving circuit and the suppression circuit are adaptively adjusted according to the environment parameters, and the power budget of the suppression circuit is improved in an extreme environment.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method for analyzing the suppression of screen noise of a liquid crystal display. Background Art

[0002] LCD screens often have screen noise during use, which seriously affects the display effect and user experience. The generation mechanism of screen noise is complex and involves multiple influencing factors. Traditional suppression methods often have problems such as unstable suppression effect and excessive power consumption. In order to effectively suppress the screen noise phenomenon, it is necessary to add a special suppression circuit to the driving circuit. However, the existing screen noise suppression technology faces two major challenges: first, the calculation complexity of the suppression circuit is high, which will significantly increase the system power consumption while achieving effective suppression; second, there is an obvious contradiction between the suppression effect and power consumption. Although increasing the suppression intensity can better eliminate the screen noise, it will also cause a sharp increase in power consumption. In addition, the use scenarios of LCD screens are complex and diverse, and the manifestation and severity of screen noise in different scenarios are also different. For example, when high-speed dynamic images are displayed, the screen noise phenomenon is more likely to occur and more obvious, requiring a stronger suppression intensity; while when static images are displayed, the screen noise phenomenon is relatively mild, and the suppression intensity can be appropriately reduced to save power consumption. Therefore, the screen noise suppression strategy needs to be able to be intelligently adjusted according to the actual display content and usage environment, and optimize power consumption while ensuring the suppression effect. How to effectively suppress the screen distortion phenomenon and ensure that the system power consumption is at a reasonable level within a limited power budget has become a key issue that needs to be urgently addressed in the field of liquid crystal display technology. Summary of the invention

[0003] The present invention provides a method for analyzing and suppressing screen noise of a liquid crystal display, which mainly comprises:

[0004] Acquire real-time working status information and properties of the liquid crystal display, and determine the current power requirement of the liquid crystal display according to the real-time working status and properties of the liquid crystal display, wherein the real-time working status information includes brightness, contrast and color saturation of the display screen, and the properties include refresh rate and resolution;

[0005] The support vector machine algorithm is used to process the real-time working status information of the LCD screen, predict the power demand of the driving circuit, and obtain the power demand curve of the driving circuit under different working conditions;

[0006] Obtain power consumption characteristic parameters of the screen noise suppression circuit, which include the required storage space and the static power consumption of the circuit. A mapping relationship between the power consumption characteristic parameters and the working intensity is established to quantify the power demand under the storage space occupancy rate and the static power consumption conditions.

[0007] Combining the real-time working status of the LCD screen, the power demand prediction curve of the driving circuit and the power consumption characteristics of the screen flicker suppression circuit, the fuzzy control algorithm is used to design the initial screen flicker suppression method, and the power budget allocation ratio between the driving circuit and the suppression circuit is dynamically adjusted;

[0008] Based on the designed initial screen flicker suppression method, the screen flicker suppression method is optimized through a neural network algorithm. On the premise of maintaining the suppression effect, the calculation process is simplified and the calculation complexity and power consumption of storage access are optimized to reduce the proportion of the suppression circuit in the entire system power consumption.

[0009] Determine whether the screen is distorted based on the real-time working status of the LCD screen. When the screen is distorted, dynamically increase the working intensity and power budget of the suppression circuit to eliminate the distorted screen. When the display is normal, reduce the suppression intensity.

[0010] The operating environment parameters of the LCD display are obtained in real time, the power distribution mode of the driving circuit and the suppression circuit is adaptively adjusted according to the environmental parameters, and the power budget of the suppression circuit is increased in extreme environments.

[0011] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0012] The invention discloses a method for suppressing and analyzing the screen noise of a liquid crystal display. In view of the screen noise problem that may occur in the working process of the liquid crystal display, the working state information of the display screen, including brightness, contrast, color saturation and attribute parameters, including refresh rate and resolution, is collected in real time, and the power demand of the driving circuit is predicted in combination with a support vector machine algorithm, and a mapping model between the power consumption characteristics and the working intensity of the screen noise suppression circuit is established. An image analysis algorithm based on edge detection is used to monitor the display screen in real time, and the screen noise phenomenon is judged by calculating the displacement and deformation degree of the edge features between adjacent frames. When the screen noise is detected, a suppression algorithm optimized by fuzzy control and neural network is used to dynamically adjust the power distribution ratio between the driving circuit and the suppression circuit, so as to achieve a balance in power consumption while ensuring the picture quality. The method can effectively solve the screen noise problem of the liquid crystal display, and at the same time realize the reasonable allocation of power resources, improve the display quality and system work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The present invention is a flow chart of a method for analyzing and suppressing screen noise of a liquid crystal display screen.

[0014] Figure 2 The schematic diagram of a method for analyzing screen noise suppression of a liquid crystal display screen according to the present invention.

[0015] Figure 3 It is another schematic diagram of a method for analyzing screen noise suppression of a liquid crystal display screen according to the present invention. DETAILED DESCRIPTION

[0016] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] like Figure 1-3 In this embodiment, a method for analyzing and suppressing screen noise of a liquid crystal display screen may specifically include:

[0018] S101. Acquire real-time working status information and properties of a liquid crystal display, and determine a current power requirement of the liquid crystal display according to the real-time working status and properties of the liquid crystal display, wherein the real-time working status information includes brightness, contrast, and color saturation of a display screen, and the properties include refresh rate and resolution.

[0019] Receive brightness data of a liquid crystal display screen collected by a multi-point brightness sensor, wherein the brightness data includes temperature values ​​at the four corners and the center of the display screen, a brightness matrix, a contrast value, and a color saturation value of a current frame of the display screen; use a Butterworth low-pass filter to obtain the average value of each parameter of the liquid crystal display screen based on the brightness data, and calculate a backlight uniformity value and a frame change rate value based on the average value; reduce the output current of the backlight drive unit when the temperature value exceeds a preset threshold, adjust the supply voltage of the drive circuit according to the frame change rate value, and compensate the signal gain multiple of the color processing unit according to the color saturation value; use a support vector regression method to establish a mapping relationship between the backlight uniformity value, the brightness matrix value, the contrast value, the color saturation value, and the frame change rate value, and obtain a power demand value of the liquid crystal display screen in the current working state.

[0020] Exemplarily, a multi-point brightness sensor is used to sample the LCD screen in real time, with a sampling interval of 50 milliseconds. Temperature sensors are arranged at the four corners and the center of the display screen to collect temperature values. The real-time brightness matrix, contrast value, and color saturation value of the current frame of the display screen are obtained from the driving circuit. The frame difference is calculated based on the two frames of image data to obtain the dynamic change of the picture, and the temperature value of each measurement point of the LCD is recorded at the same time. The current operating parameters are read from the LCD display driver chip, including the driving voltage, backlight current, and refresh timing. The collected multi-frame data is filtered using a Butterworth low-pass filter to remove measurement noise, and the average value of each parameter of the LCD screen is obtained. The backlight uniformity value is calculated from the multi-point brightness sensor data, and the frame change rate is calculated based on the dynamic change of two adjacent frames. If the value of any temperature measurement point exceeds 85 degrees Celsius, the output current of the backlight drive unit is reduced according to the preset curve; if the frame change rate is greater than 0.5, the drive circuit power supply voltage is increased by 0.2 volts through the voltage stabilizer; if the color saturation of the display screen is greater than 0.8, the signal gain multiple of the color processing unit is adjusted to 0.9 to dynamically compensate for the saturation. Support vector regression is used to establish a mapping between display parameters and power consumption, and the current resolution parameters and refresh rate parameters of the liquid crystal display screen are obtained from the driver chip. Combined with the backlight uniformity value, brightness matrix value, contrast value, color saturation value, and frame change rate value, the power demand value of the display screen in the current working state is calculated and output. The liquid crystal display screen adopts a multi-point brightness sensor arrangement scheme to monitor the backlight uniformity of the display panel. Photoresistors are arranged in the four corners and the center area of ​​the display screen. The sampling time interval is set to 50 milliseconds. The brightness value detected by the photoresistor is between 0-255, and the brightness value of the center area is usually about 15% higher than that of the corner area. When the brightness value of the corner area is lower than 30% of the center area, it indicates that there is a problem with the backlight uniformity. Temperature sensors are also arranged at the four corners and the center of the display screen to collect temperature data in real time. The normal operating temperature of the LCD screen is between 25 and 65 degrees Celsius. Temperatures exceeding 85 degrees Celsius will cause the liquid crystal molecules to be arranged in disorder. The dynamic change of the display screen reflects the response speed of the liquid crystal molecules. It is obtained by calculating the difference between two adjacent frames of images. The brightness value of each pixel is between 0 and 255, and the calculated frame change rate is between 0 and 1. The frame change rate of 0.5 means that the average brightness value of each pixel changes by more than 127. The Butterworth low-pass filter processes the collected data, and the cutoff frequency is set to 100 Hz, which can effectively remove high-frequency noise while retaining the rapid change characteristics of the display parameters. The driving voltage of the LCD screen is closely related to the refresh rate. The driving voltage is 3.3 volts at a refresh rate of 60 Hz. When the frame change rate exceeds 0.5, the driving voltage is increased to 3.5 volts to speed up the response speed of the liquid crystal molecules.Color saturation is between 0 and 1. Saturation 0.8 means that at least one of the three RGB components reaches 80% of the maximum value. By adjusting the signal gain multiple to 0.9, display distortion caused by color oversaturation can be avoided. The support vector regression algorithm establishes a mapping relationship between display parameters and power consumption. The training samples include display parameters and corresponding power consumption values ​​under different working conditions. For example, the power consumption is 15 watts when the resolution is 1920x1080, the refresh rate is 60 Hz, the backlight uniformity is 0.85, the average brightness value is 200, the contrast ratio is 2000 to 1, the color saturation is 0.7, and the frame change rate is 0.3. Each display parameter has a different degree of influence on power consumption, among which the backlight brightness and refresh rate have the greatest influence. For every 10% increase in backlight brightness, the power consumption increases by 8%. When the refresh rate increases from 60 Hz to 120 Hz, the power consumption increases by 35%. In a high-temperature working environment, the power consumption can be effectively reduced by reducing the backlight brightness. When the temperature reaches 85 degrees Celsius, reducing the backlight brightness to 70% of the normal value can reduce the power consumption by 25%.

[0021] The panel attribute parameters of the LCD screen when it is working in real time are collected, and the panel attribute parameters are compared with the preset standard values ​​to obtain the deviation value; the refresh rate and resolution attributes of the LCD panel are used as input conditions for power demand calculation, and linear interpolation calculation is performed in combination with the real-time working state deviation value and the panel attribute parameters to obtain the real-time power demand value of the LCD screen in the current working state.

[0022] The brightness value, contrast value and color saturation value collected by the liquid crystal display screen sensor are obtained, and a difference operation is performed between the values ​​and the reference values ​​in the preset standard parameter table to obtain a brightness deviation value, a contrast deviation value and a color saturation deviation value; the refresh rate value and the display resolution parameter read by the liquid crystal display screen driver chip are normalized to obtain a basic power demand coefficient; according to the brightness deviation value, contrast deviation value and color saturation deviation value, a power compensation coefficient is generated by using cubic spline interpolation; the power compensation coefficient and the basic power demand coefficient are combined by a gradient boosting tree to obtain a power demand value of the liquid crystal display screen.

[0023] Exemplarily, by arranging photoelectric sensors at the center and four corners of the liquid crystal display screen, the brightness value is collected every 20 milliseconds, the contrast value is obtained from the display driving circuit, and the color saturation value is collected every 50 milliseconds using a color sensor, wherein the brightness value is in the range of 0 to 255, the contrast value is in the range of 500 to 3000, and the color saturation value is in the range of 0 to 1. The difference between the brightness value, the contrast value, and the color saturation value and the reference value stored in the preset standard parameter table is calculated to obtain the real-time deviation value of the three parameters. The current working timing parameters are read from the liquid crystal display driver chip to obtain the real-time refresh rate value of the liquid crystal display screen, including four gears of 30 Hz, 60 Hz, 75 Hz, and 120 Hz. The current display resolution parameter is collected from the image processing unit, and the refresh rate value is multiplied by the display resolution parameter, and normalized with 1920 multiplied by 1080 pixel resolution and 60 Hz refresh rate as the reference point to generate the basic power demand coefficient. According to the 32 groups of power values ​​under standard working conditions recorded in the power demand mapping table, each group of data includes display parameter values ​​and corresponding reference power, cubic spline interpolation is used to process the brightness deviation value, contrast deviation value, and color saturation deviation value, and the power compensation coefficient under real-time state is calculated according to the interpolation curve, and the power compensation coefficient is limited to the range of 0.5 to 1.5. The standard power reference value and the power compensation coefficient are combined and calculated using a gradient boosting tree, and the power demand value of the liquid crystal display screen under real-time working state is calculated by inputting the brightness deviation value, contrast deviation value, color saturation deviation value, and basic power demand coefficient, and the power demand value is calculated in watts. The brightness sensor arrangement of the liquid crystal display screen adopts a five-point sampling scheme, and photoresistors are placed at the center point and four corners of the screen to monitor the brightness uniformity of the entire display area. Sampling is performed once every 20 milliseconds. The brightness value detected by the photoresistor is between 0 and 255, and the center area is usually 10% to 15% higher than the corner area. When the brightness of the corner area is 25% lower than that of the center area, it indicates that the backlight uniformity is abnormal. The color sensor has a sampling interval of 50 milliseconds and measures the color saturation of the display screen, with a value between 0 and 1. A saturation of 0.8 means that at least one of the three RGB components reaches 80% of the maximum value. The contrast parameter output by the display driver circuit reflects the brightness ratio between the brightest area and the darkest area of ​​the screen. The contrast is between 500 and 3000, and is usually maintained in the range of 1000 to 2000 under standard working conditions. Too high a contrast will cause loss of details in the dark part of the screen, and too low a contrast will cause insufficient sense of hierarchy in the screen. The preset standard parameter table stores standard parameter values ​​for different scenarios, including the optimal parameter combination for typical application scenarios such as office mode, game mode, and video mode.The LCD screen supports multiple refresh rates, including 30 Hz, 60 Hz, 75 Hz, and 120 Hz. The resolution covers 1280 times 720, 1920 times 1080, 2560 times 1440 and other specifications. The 1920 times 1080 resolution and 60 Hz refresh rate are used as the benchmark configuration. Under the benchmark configuration, the standard power of the display is 10 watts. When the resolution is increased to 2560 times 1440, the basic power demand coefficient increases to 1.5. When the refresh rate is increased to 120 Hz, the basic power demand coefficient increases to 1.8. The power demand mapping table records 32 sets of standard power values ​​under different working conditions. The power compensation coefficient under any working condition is calculated by cubic spline interpolation. The compensation coefficient is limited to the range of 0.5 to 1.5 to avoid over-compensation. In the standard display mode, when the brightness is set to 200, the contrast is 1500, and the color saturation is 0.7, the power compensation coefficient is 1.0. When the brightness is increased to 250, the contrast is increased to 2000, and the color saturation is increased to 0.9, the power compensation coefficient increases to 1.3. The gradient boosting tree calculates the actual power demand under the current working state by inputting the real-time deviation value of the display parameters and the basic power demand coefficient, accurately reflecting the dynamic power consumption changes of the display in different working modes.

[0024] S102, using a support vector machine algorithm to process the real-time working state information of the liquid crystal display screen, predicting the power demand of the driving circuit, and obtaining a power demand curve of the driving circuit under different working states.

[0025] A circuit integrated measurement unit is used to obtain working status data of a liquid crystal display screen, wherein the working status data includes a brightness value and a contrast value output by a display driver chip, a temperature value collected by a temperature sensor, and a voltage value on a current sampling resistor; a support vector machine model is established according to the working status data, and a nonlinear mapping is performed on the working status data through a Gaussian kernel function to obtain a power prediction value of the display screen; a reference mapping table is established for the power prediction value, and a cubic spline curve is used to continuously interpolate adjacent measurement points in the reference mapping table to obtain a power demand curve; and according to a power value corresponding to the current working state in the power demand curve, a bus voltage regulator is used to adjust the power supply voltage of the drive circuit to obtain an output limit value of the power supply voltage.

[0026] Exemplarily, the working state data of the liquid crystal display screen is collected every 50 milliseconds through the measurement unit integrated with the circuit, the brightness value, contrast value, color saturation value, refresh frequency value, and resolution value are read from the display driver chip, the temperature sensors at the four corners and the center of the back panel of the display screen collect the working temperature value, the voltage value on the current sampling resistor is recorded to convert the real-time current value, and the dynamic change rate of the picture is calculated based on the pixel difference between two adjacent frames of the image to form a complete working state data set. The collected working state data is modeled using a support vector machine, and the input features include display brightness value, contrast value, color saturation value, refresh frequency value, resolution value, temperature value, current value, and dynamic change rate. The training data label is the measured power consumption value. The data is nonlinearly mapped through the Gaussian kernel function to establish a mapping relationship between working parameters and power requirements, and the power prediction value of the driving circuit under different working states is output. A reference mapping table is established for 32 groups of power measurement data under typical working conditions. The values ​​between adjacent measurement points are continuously interpolated using a cubic spline curve. The power values ​​predicted by the support vector machine are corrected based on the interpolation function to generate a continuous and smooth power demand curve, which reflects the real-time power demand of the liquid crystal display screen under any working state. The power supply configuration of the drive circuit is guided by the power demand curve, the power value corresponding to the current working state is queried from the curve, the power supply voltage of the drive circuit is dynamically adjusted by the bus voltage regulator, and the output limit of the power supply voltage is set according to the predicted power value to realize the real-time configuration of the power resources of the liquid crystal display screen. The collection process of the working state data of the liquid crystal display screen runs through the entire operation cycle, and the data is sampled every 50 milliseconds by the measurement unit integrated in the circuit to realize real-time monitoring of the display parameters. The brightness value output by the display driver chip ranges from 0 to 255, the contrast value ranges from 500 to 3000, and the color saturation value ranges from 0 to 1. When the display content switches from static text to dynamic video, the brightness value increases from 160 to 200, the contrast value increases from 1000 to 1800, the color saturation value increases from 0.6 to 0.8, and the refresh rate increases from 60 Hz to 120 Hz. The temperature sensors are arranged at the four corners and the center of the back panel of the display screen to form a five-point temperature monitoring network. The normal operating temperature is between 25 and 65 degrees Celsius. When the difference between the corner temperature and the center temperature exceeds 15 degrees Celsius, it indicates uneven heat dissipation. The current sampling resistor has a resistance of 0.1 ohms. When the measured voltage value is 0.5 volts, the corresponding working current is 5 amperes. The dynamic change rate of the picture is obtained by calculating the pixel difference between two adjacent frames of the image. When switching from a static interface to video playback, the dynamic change rate increases from 0.1 to 0.6. The support vector machine modeling uses a Gaussian kernel function for nonlinear mapping. The kernel function parameter is set to 0.01 to fit the relationship between display parameters and power consumption.At a resolution of 1920 x 1080, when the display brightness value is 200, the contrast value is 1500, the color saturation value is 0.7, and the refresh rate is 60 Hz, the measured power consumption is 15 watts, and the predicted power consumption is 15.2 watts, with a prediction error within 5%. When the resolution is increased to 2560 x 1440 and other parameters remain unchanged, the power consumption rises to 22 watts. The benchmark mapping table records 32 sets of power measurement data under typical working conditions, and uses cubic spline curves for interpolation to ensure that the power prediction curve transitions smoothly between each measurement point. In the game scenario, when the screen brightness value is 220, the contrast value is 2000, the color saturation value is 0.8, and the refresh rate is 120 Hz, the interpolation calculation results in a power consumption of 28 watts. The bus voltage regulator dynamically adjusts the supply voltage according to the predicted power consumption. When the predicted power consumption is 15 watts, the supply voltage is set to 12 volts, and the output current is limited to 1.25 amperes. When the predicted power consumption rises to 28 watts, the supply voltage is increased to 14 volts and the current limit is increased to 2 amps, achieving dynamic allocation of power resources.

[0027] S103, obtaining power consumption characteristic parameters of the screen noise suppression circuit, the power consumption characteristic parameters including the required storage space and the static power consumption of the circuit, and establishing a mapping relationship between the power consumption characteristic parameters and the working intensity to quantify the storage space occupancy rate and the power demand under the static power consumption conditions.

[0028] The storage unit usage, power supply voltage, working current and temperature values ​​in the screen flicker suppression circuit are collected, and a cache hit rate value is calculated based on the collected data; a random forest model is constructed based on the storage space occupancy, cache hit rate, temperature value, voltage-current product and suppression task processing volume, and a power consumption prediction value is obtained through the random forest model; the gear to which the suppression task processing volume value belongs is determined, and if the processing volume value is in a light load interval, a first gear value is obtained; if the processing volume value is in a medium load interval, a second gear value is obtained; if the processing volume value is in a heavy load interval, a third gear value is obtained; and if the processing volume value is in a full load interval, a fourth gear value is obtained; historical record points are constructed using the gear values ​​and the power consumption prediction value, and a working gear and power demand mapping function is obtained by fitting the historical record points, and the power demand value corresponding to any workload gear is obtained according to the mapping function.

[0029] Exemplarily, real-time operation data is collected from the screen flicker suppression circuit, and the usage of the storage unit is read every 100 milliseconds, including the number of bytes occupied by the memory space, the number of bytes used by the cache, and the size of the memory mapping table. The power supply voltage value and the working current value are collected every 10 milliseconds through the voltage and current sampler. The temperature detector is arranged in the data processing unit, storage control unit, and timing control unit of the circuit to collect the temperature value every 50 milliseconds, and the cache hit rate value is calculated based on a 10-minute statistical period. The power consumption characteristic parameters of the screen flicker suppression circuit are modeled using a random forest algorithm, and a 5-dimensional feature vector including storage space occupancy, cache hit rate, temperature value, voltage-current product, and suppression task processing volume is constructed. The average static power consumption within the 10-minute sampling interval is recorded as the training label value, and the feature vector is input to obtain the corresponding power consumption prediction value. The suppression task processing volume is divided into 4 gears according to the number of data bytes processed per minute, corresponding to the light load range of 0 to 100 megabytes, the medium load range of 100 to 500 megabytes, the heavy load range of 500 to 1000 megabytes, and the full load range of 1000 megabytes or more. The gear is determined according to the current collected processing volume value, and the gear value and the power consumption value predicted by the random forest form a corresponding point. The power demand value under different workload gears is recorded, and the power demand curve is drawn based on 32 historical record points. The mapping function of the working gear and the power demand is obtained by cubic curve fitting. The corresponding power demand value is calculated for any input workload gear, and the power demand value is limited to the range of 0 to 100 watts. The storage unit of the screen flower suppression circuit includes three parts: memory, cache and mapping table. The memory mainly stores the image data to be processed, and occupies about 500 megabytes in the typical working state. The cache is used to temporarily store the processing results, occupying about 50 megabytes. The mapping table records the data address information, occupying about 10 megabytes. The sampler measures the supply voltage and operating current every 10 milliseconds. During normal operation, the voltage is 3.3 volts and the current varies between 0.5 and 2 amperes. Temperature detectors are arranged in three key units. The temperature of the data processing unit is between 45 and 65 degrees Celsius, the temperature of the storage control unit is between 40 and 55 degrees Celsius, and the temperature of the timing control unit is between 35 and 50 degrees Celsius. The random forest model training uses a 5-dimensional feature vector. When the storage space occupancy rate is 60%, the cache hit rate is 85%, the average temperature is 50 degrees Celsius, the power supply is 5 watts, and the processing volume is 300 megabytes per minute, the static power consumption is measured to be 3 watts. When the value of the feature vector changes, the power consumption changes accordingly. For every 10% increase in storage occupancy, the power consumption increases by 0.2 watts, for every 5% decrease in cache hit rate, the power consumption increases by 0.3 watts, and for every 5 degrees Celsius increase in temperature, the power consumption increases by 0.4 watts. The workload division of the suppression task reflects the processing pressure of the circuit. In the light load range of 0 to 100 megabytes per minute, it mainly handles occasional screen flickering of static interfaces, and the power consumption is kept below 5 watts.The medium load range of 100 to 500 megabytes per minute corresponds to video playback scenarios, and the power consumption is between 5 and 15 watts. The heavy load range of 500 to 1000 megabytes per minute corresponds to game screens, and the power consumption is between 15 and 30 watts. The full load range of more than 1000 megabytes corresponds to professional graphics processing, and the power consumption exceeds 30 watts. The 32 historical record points of the power demand curve cover different working scenarios. In the light load range, the power consumption increases by 1 watt for every increase of 50 megabytes per minute of processing. In the medium load range, the power consumption increases by 2 watts for every increase of 100 megabytes per minute. In the heavy load range, the power consumption increases by 5 watts for every increase of 200 megabytes per minute. In the full load range, the power consumption increases by 15 watts for every increase of 500 megabytes per minute. The mapping function after cubic curve fitting accurately reflects the nonlinear relationship between workload and power consumption, and the prediction error is controlled within 10%, realizing the accurate quantification of the power demand of the screen suppression circuit.

[0030] S104. Combining the real-time working status of the LCD display, the power demand prediction curve of the driving circuit, and the power consumption characteristics of the screen flicker suppression circuit, a fuzzy control algorithm is used to design an initial screen flicker suppression method, and the power budget allocation ratio between the driving circuit and the suppression circuit is dynamically adjusted.

[0031] A state evaluation vector is constructed according to the screen brightness value, contrast value and color saturation value obtained in real time; a triangular membership function is established according to the state evaluation vector, a rule mapping table is generated for the membership function, and the rule reasoning result is obtained by the centroid method; the weighted average method is used to calculate the weight coefficient of the working state parameter corresponding to the rule reasoning result, and the driving circuit power prediction curve is segmented according to the weight coefficient; if the product of the weight coefficient and the quality score is greater than the preset threshold, the driving circuit power share is increased, and the power budget allocation ratio is updated by the exponential sliding average method.

[0032] Exemplarily, the screen brightness value, contrast value, and color saturation value are read from the display screen data acquisition unit every 20 milliseconds, the brightness value is standardized in the range of 0 to 255, the contrast value is standardized in the range of 500 to 3000, and the color saturation value is standardized in the range of 0 to 1. The storage occupancy value, static power consumption value, and real-time processing value of the screen noise suppression circuit are collected to construct a state evaluation vector containing 7-dimensional parameters, and each parameter is mapped to the range of 0 to 1. A triangular membership function is constructed based on the state evaluation vector, and three levels of low, medium, and high membership are set for the screen working state parameters, three levels of energy-saving standard performance membership are set for the driving circuit power prediction value, and three levels of light load, medium load, and heavy load membership are set for the suppression circuit power consumption characteristics. A rule mapping table with 27 combinations is generated, and the rule reasoning result is calculated by the centroid method. The weight coefficient of the working state parameter is calculated by weighted average method. The power prediction curve of the driving circuit is segmented based on the weight coefficient. The total power is divided into three fixed ratios of 4:6, 5:5, and 6:4 to the driving circuit and the suppression circuit. The edge sharpness, number of noise points, and color uniformity of the displayed image are calculated to obtain a quality score in percentage. The proportion of power budget allocation is updated online using exponential sliding average. The sliding window length is set to 100 milliseconds. The product of the weight coefficient and the quality score is used as the evaluation index. When the index value is greater than the threshold of 0.8, the proportion of driving circuit power is increased. When the index value is less than the threshold of 0.6, the proportion of suppression circuit power is increased. The adjustment step size is limited to 5% each time. The standardized processing of LCD display parameter acquisition reflects the characteristics of different working states. When displaying static text, the brightness value is about 160, the contrast is about 1000, and the color saturation is about 0.6. After standardization, they correspond to 0.63, 0.33, and 0.6 respectively. When switching to video playback, the parameters change to brightness value 200, contrast 1800, color saturation 0.8, and the standardized values ​​are 0.78, 0.6, and 0.8, respectively, reflecting the impact of different display contents on screen parameters. The design of the membership function reflects the core idea of ​​fuzzy control. Taking the brightness parameter as an example, the low-range interval covers 0 to 0.4, the mid-range interval covers 0.3 to 0.7, and the high-range interval covers 0.6 to 1.0, achieving a smooth transition between gears. When the brightness standardization value is 0.35, it belongs to both low and mid-range, with memberships of 0.7 and 0.3, respectively. The 27 rule combinations cover various working states. For example, when the combination of low brightness, energy-saving mode, and light load state, the drive circuit is allocated 40% power and the suppression circuit is allocated 60% power. The weight coefficient calculation reflects the importance of different parameters. When the game screen is displayed, the weight of the dynamic parameter increases, and the weight ratio of the brightness value, contrast value, and color saturation value becomes 3:2:1. In the static office scene, the weight ratio of the three parameters is changed to 1:2:3. The power prediction curve is divided into three sections: high, medium and low according to the weight coefficient. The power levels are subdivided in the interval with large weight to improve the allocation accuracy.The quality score is based on the image processing results. The edge sharpness is calculated by the Laplace operator. A score above 75 indicates good clarity. The number of noise points is obtained by local variance statistics. It is qualified if it is less than 1% of the image area. The color uniformity is calculated based on the deviation of adjacent pixels. The standard deviation is less than 5% for normal. The exponential sliding average gives more weight to the recent data. Within the window length of 100 milliseconds, the weight of the newly sampled data is 0.9, and the weight of the historical data decays to 0.1. When the quality score is 85 points and the weight coefficient is 0.9, the index value is 0.85, which is greater than the threshold of 0.8, and the proportion of driving circuit power increases by 5%. When the quality score drops to 70 points and the weight coefficient is 0.8, the index value is 0.56, which is less than the threshold of 0.6, and the proportion of suppression circuit power increases by 5%, achieving a dynamic balance of power distribution.

[0033] S105. Based on the designed initial flower screen suppression method, the flower screen suppression method is optimized through a neural network algorithm. On the premise of maintaining the suppression effect, the calculation process is simplified and the calculation complexity and power consumption of storage access are optimized to reduce the proportion of the suppression circuit in the overall system power consumption.

[0034] The original flower screen image and the suppression result image output by the image data acquisition unit are obtained, and the operation type and data volume of the operation unit are recorded according to the original flower screen image and the suppression result image; the operation type and data volume of the operation unit are modeled by using a convolutional neural network, and the contribution of the operation branch is obtained by calculating the loss function, and a pruning operation is performed on the operation branch whose contribution is less than the loss function threshold; for the operation branch after the pruning operation, a local cache is used to store the continuous access data of the operation unit, a cache mapping table is established according to the continuous access data, and a fixed-point conversion is performed on the floating-point operation of the operation unit; the voltage and current values ​​of the operation unit are collected by a power consumption detection circuit, the power consumption curve is calculated according to the voltage and current values, and the image quality score is obtained based on the image edge sharpness and the number of noise points.

[0035] Exemplarily, the original screen image and the suppression result image are obtained from the image data acquisition unit every 10 milliseconds, the operation type, input data volume, and output data volume of each operation unit in the screen suppression method are recorded, the number of operations of the calculation unit within a 50-millisecond statistical period is monitored, the number of read and write times and read and write addresses of the storage unit are collected, and an optimized data set containing operation logs, access records, image edge sharpness values, image noise number, and color uniformity values ​​is constructed. A convolutional neural network is used to model the operation process, the input layer sets 16 channels to receive operation features, the middle layer sets 4 layers of convolution to extract the correlation of operation features, and the output layer corresponds to the optimized operation process. The loss function combines the three indicators of the number of operations, the number of accesses, and the image quality score. When the contribution of a certain operation branch is less than 5% of the total loss function, the branch is pruned. For the pruned operation flow chart, a local cache with a capacity of 512 bytes is used to store data blocks that have been accessed for 8 consecutive times, a cache mapping table is established for data with an access interval of less than 20 milliseconds, and the operation operations with the same input data are merged. The floating-point addition, subtraction, multiplication and division operations are converted into fixed-point operations, and the intermediate result data is quantized to 8 bits. The power consumption detection circuit collects voltage and current values ​​every 1 millisecond, and calculates the power consumption curves of the two methods before and after optimization. It may include transmitting the collected voltage and current data to the data processing module, and obtaining the power consumption data within a period of time through integral operation; according to the set time interval, the continuous power consumption data is divided into multiple time windows, and the average power consumption in each time window is used as the power consumption value at that time point; for the two methods before and after optimization, the power consumption curves composed of the power consumption values ​​at each time point are drawn respectively, and the power consumption characteristics changing with time are intuitively displayed. The image edge sharpness is calculated based on the Laplace operator, the number of noise points is counted by local variance, and the color uniformity is calculated using the color difference of adjacent pixels. If the image quality score drops by more than 5%, it will fall back to the last optimization result, and the optimization process will be re-executed until the score changes by less than 1%. The operation process optimization of the flower screen suppression method involves data collection and analysis at multiple levels. When processing an image with a resolution of 1920 by 1080, the original algorithm includes three main operation units: edge detection, noise filtering, and color correction. The edge detection unit processes 2.07 million pixels every 10 milliseconds, performs 4 convolution operations, and the cumulative number of operations reaches 8.28 million times. The noise filtering unit samples each pixel at 9 points, with an operation count of 18.63 million times. The color correction unit performs three-channel processing, with an operation count of 6.21 million times. Within a 50-millisecond statistical cycle, the storage unit reads and writes 33.12 million times. The convolutional neural network uses a multi-layer structure to optimize the operation process. The 16 channels of the input layer correspond to different types of operation features, including the number of operations, data read and write volume, number of intermediate results, storage address span, and other information. The four convolutional layers in the middle extract feature correlations through a 3x3 convolution kernel to identify redundant parts in the operation chain.The weight of the number of operations in the loss function is 0.4, the weight of the number of accesses is 0.4, and the weight of the quality score is 0.2. When the contribution rate of the secondary convolution operation in edge detection to the total loss is less than 5%, the operation link is cut off. The optimized operation process adopts data reuse. The 512-byte local cache is divided into 64 data blocks, and each block caches the data accessed 8 times in a row. In video image processing, the color correction parameters of adjacent pixels are similar. The cache mapping table records the data location accessed within the last 20 milliseconds, and the data reuse rate is increased to 85%. After the floating-point operation is converted to fixed-point operation, the number of multiplication operations is reduced by 40%, and the number of addition and subtraction operations is reduced by 25%. 8-bit fixed-point quantization reduces the data accuracy to 256 levels and the storage space is reduced by 50%. The power consumption detection circuit uses a current sampling resistor to obtain the real-time current value. When the original algorithm processes 1080p images, the peak power consumption curve reaches 15 watts. The peak power consumption of the optimized algorithm is reduced to 9 watts. Image quality evaluation uses multiple indicators. Edge sharpness is calculated using the Laplace operator. The original algorithm scored 92 points, and after optimization it was 89 points, a decrease of 3.3%. The number of noise points was reduced from 0.8% of the original image to 0.5%. The color uniformity score was based on the standard deviation of the color difference of adjacent pixels, which was reduced from 95 points to 93 points. After multiple rounds of optimization iterations, the change in quality score was stabilized within 0.8%, and power consumption was reduced by 40%.

[0036] S106. Determine whether a screen distortion phenomenon occurs according to the real-time working state of the liquid crystal display screen. When a screen distortion phenomenon is detected, dynamically increase the working intensity and power budget of the suppression circuit to eliminate the screen distortion. When the display is normal, reduce the suppression intensity.

[0037] Extract edge contour feature values ​​according to display screen image data, calculate edge difference data of adjacent image frames through the edge contour feature values; obtain the edge difference data, calculate noise density values ​​by using a sliding window, and construct a feature vector according to the noise density values ​​and color saturation values; perform support vector machine classification operation on the feature vector, and obtain a screen flower level determination result through the classification operation; adjust the operating frequency parameter, operation precision parameter and storage bandwidth parameter according to the screen flower level determination result, and if the edge difference data in a continuous monitoring period is lower than a preset threshold value, reduce the operating frequency parameter, the operation precision parameter and the storage bandwidth parameter according to a preset gradient period.

[0038] Exemplarily, a display screen sampler collects one frame of image data every 10 milliseconds, uses the Sobel operator to extract the edge contour of the image, calculates the edge pixel difference between two adjacent frames of the image, uses a 16 by 16 pixel sliding window to calculate the local variance, counts the noise distribution density in each window, and combines the maximum saturation values ​​of the three channels of red, green and blue to construct a 7-dimensional feature vector including edge difference, noise density, and color saturation. A support vector machine with a radial basis kernel function is used to classify the feature vectors, and a three-dimensional decision boundary is constructed by using an edge difference threshold of 0.3, a noise density threshold of 0.2, and a color saturation threshold of 0.8. The degree of screen distortion is divided into three levels: mild, moderate, and severe. The screen distortion level is quantified based on the image quality score, with mild corresponding to a score of 70 to 85 points, moderate corresponding to 50 to 70 points, and severe corresponding to less than 50 points. In the case of mild screen distortion, the operating frequency of the suppression circuit is increased to 1.2 times the reference frequency, the calculation accuracy is maintained at 16-bit fixed-point, and the storage bandwidth is increased to 1.3 times the reference bandwidth; in the case of moderate screen distortion, the operating frequency is increased to 1.5 times, the calculation accuracy is increased to 24-bit fixed-point, and the storage bandwidth is increased to 1.6 times; in the case of severe screen distortion, the operating frequency is increased to 2 times, the calculation accuracy is increased to 32-bit fixed-point, and the storage bandwidth is increased to 2 times. Use a 100-millisecond sliding time window to monitor the display image quality. When the edge difference value is less than 0.2, the noise density is less than 0.1, and the color saturation is less than 0.7 within 5 consecutive monitoring cycles, the operating frequency is reduced by 0.1 times, the calculation accuracy is reduced by 4 bits, and the storage bandwidth is reduced by 0.1 times with a gradual cycle of every 20 milliseconds until all parameters return to the reference level. The detection of LCD screen noise involves the collection and analysis of multiple image features. The image data collected every 10 milliseconds contains 1920 by 1080 pixels, and each pixel contains three 8-bit color channels of red, green and blue. The Sobel operator extracts the image edges in the horizontal and vertical directions. The edge difference value of two adjacent frames is below 0.1 in normal display, and the difference value rises to 0.2 to 0.3 when there is mild noise, and exceeds 0.4 when there is severe noise. When the 16 by 16 pixel sliding window moves on the image, the standard deviation of the pixel value in the window reflects the degree of noise. The standard deviation is within 5 in normal display, rises to 10 to 15 when there is mild noise, and exceeds 20 when there is severe noise. The radial basis kernel function of the support vector machine maps the 7-dimensional feature vector and constructs a nonlinear classifier. In the game screen display scenario, when the edge difference value is 0.25, the noise density is 0.15, and the color saturation is 0.75, it is judged as a mild noise screen. In the video playback scenario, when the edge difference value is 0.35, the noise density is 0.25, and the color saturation is 0.85, it is judged as moderate screen distortion. In the graphics processing scenario, when the edge difference value is 0.45, the noise density is 0.35, and the color saturation is 0.95, it is judged as severe screen distortion. The image quality score is based on a percentage system, with edge clarity accounting for 30 points, noise control accounting for 40 points, and color uniformity accounting for 30 points.The working parameters of the suppression circuit are dynamically adjusted according to the degree of screen distortion. In the baseline working state, the operating frequency is 200 MHz, 16-bit fixed-point operation is used, and the storage bandwidth is 4 GBytes per second. When a mild screen distortion is detected, the frequency is increased to 240 MHz, 16-bit fixed-point operation is maintained, and the bandwidth is increased to 5.2 GBytes per second. When the screen distortion is moderate, the frequency is increased to 300 MHz, the precision is increased to 24 bits, and the bandwidth is increased to 6.4 GBytes per second. When the screen distortion is severe, the frequency is increased to 400 MHz, the precision is increased to 32 bits, and the bandwidth is increased to 8 GBytes per second. The parameter adjustment after the display returns to normal adopts a smooth transition method. When the edge difference value, noise density, and color saturation of 5 consecutive samples in the 100-ms monitoring window return to normal, the frequency is reduced by 20 MHz every 20 milliseconds, the calculation precision is reduced by 4 bits, and the bandwidth is reduced by 0.4 GBytes per second. It usually takes 200 milliseconds for all parameters to return to the baseline level, avoiding screen flickering caused by parameter mutations. This gradual adjustment achieves a steady reduction in power consumption while ensuring display quality.

[0039] Monitor the display screen of the LCD screen in real time, and use the image analysis method based on edge detection to extract the position and shape characteristics of the edge of the object in the picture; by calculating the displacement and deformation degree of the edge features between adjacent frames, determine whether the picture has broken, ghosting or other screen-distorting phenomena. If the screen-distorting phenomenon is detected, dynamically adjust the working intensity parameters of the suppression circuit according to the quantitative index of the screen-distorting degree, and increase its allocation ratio in the overall power budget accordingly, until the quantitative index of the screen-distorting phenomenon is reduced to below the preset threshold.

[0040] The image data of the liquid crystal display screen is obtained through an image processing unit, and the edge contour is extracted by using the Laplace operator according to the image data, and the coordinate value and curvature value of the feature point are obtained for the edge contour; the displacement vector between consecutive frames is calculated by using the pyramid optical flow method according to the feature point, and if the length of the displacement vector exceeds a preset threshold, it is determined to be a ghosting area, and if the displacement direction deviation is less than a preset angle, it is determined to be the same ghosting group; the pixel ratio of the ghosting area is counted to obtain a ghosting index, and the number of curvature discontinuity points is calculated for the edge contour to obtain a fracture index, and a quantitative value of the degree of screen noise is obtained according to the ghosting index and the fracture index according to the preset weights; an average value is calculated by using a sliding window according to the quantified value, and if the average value exceeds a first preset threshold, the power budget of the suppression circuit is increased, and if the average value is lower than a second preset threshold, the power budget is reduced to a reference level.

[0041] Exemplarily, the image processing unit collects image data of the LCD screen every 8.3 milliseconds, synchronized with the 120 Hz refresh rate of the display, uses the Laplace operator to extract the edge contour of the object in the picture, uses the cubic spline curve to fit the edge contour, selects feature points at the inflection point of the curve, selects 5 feature points for each curve segment, records the coordinate values ​​and curvature values ​​of the feature points, and constructs an edge feature descriptor containing position and shape information. The pyramid optical flow method is used to calculate the displacement vector of the feature points between consecutive frames. The displacement vector length exceeds 5 pixels and is determined to be a ghosting area. The displacement direction deviation is less than 15 degrees and is determined to be the same ghosting group. The percentage of pixels in the ghosting area is counted to obtain the ghosting index, and the number of curvature discontinuities of the edge curve is calculated to obtain the fracture index. The ghosting index and the fracture index are weighted according to a weight of 0.6 to 0.4 to generate a quantitative value of the degree of screen noise. The screen noise level is divided according to the quantization value, 0 to 0.3 is mild, 0.3 to 0.6 is moderate, and 0.6 or above is severe. When the screen noise is mild, the operating frequency of the suppression circuit is increased from the base 200 MHz to 240 MHz, and the data cache capacity is maintained at 512 kilobytes; when the screen noise is moderate, the frequency is increased to 300 MHz, the cache is expanded to 1024 kilobytes; when the screen noise is severe, the frequency is increased to 400 MHz, the cache is expanded to 2048 kilobytes, and the operating voltage is increased from 1.2 volts to 1.5 volts. The screen noise quantization value is monitored using a 50-millisecond sliding window, and the window slides forward every 10 milliseconds. When the average quantization value of three consecutive windows exceeds 0.3, the suppression circuit power budget is increased by 20% of the base value, and the power budget is increased by 20% for each quantization value exceeding 0.3; when the average quantization value of five consecutive windows is lower than 0.2, the power budget is reduced by 5% every 10 milliseconds until it returns to the base level. The detection of LCD screen noise involves edge feature extraction and motion analysis. When displaying high-definition video, the sampling interval of 8.3 milliseconds is synchronized with the refresh rate of 120 Hz to ensure that every frame of the picture changes is captured. The Laplace operator performs a second-order derivative operation on the image to extract the edge features of the object contour. For a 1920 by 1080 resolution picture, the edge extraction processing time is controlled within 2 milliseconds. The cubic spline curve fitting connects the discrete edge points into a smooth curve, and selects feature points at positions where the curvature changes by more than 30 degrees. A typical rectangular object contour will select 4 corner points and the midpoints of 4 edges, a total of 8 feature points. The pyramid optical flow method matches feature points by constructing a 4-layer image pyramid and calculates the displacement vector from top to bottom. Under normal display conditions, the displacement of feature points between adjacent frames is less than 2 pixels, and the displacement reaches 5 to 8 pixels when slight ghosting occurs, and exceeds 10 pixels when severe ghosting occurs. The standard deviation of the displacement direction is less than 5 degrees in normal display and reaches more than 20 degrees when ghosting occurs. Statistics on the ghosting area show that when the screen is mildly distorted, the affected area accounts for 10% of the screen area, when it is moderate, it accounts for 25%, and when it is severe, it exceeds 40%.The breakage index reflects the continuity of the edge curve. By detecting the curvature mutation point statistics, the number of breakage points per 100 edge pixels is less than 2 in normal display, and increases to 5 to 10 when the screen is distorted. The working parameters of the suppression circuit are dynamically adjusted according to the degree of screen distortion. Under the baseline state, the operating frequency of 200 MHz is more than enough to process 1080p video, and the 512 kilobyte cache can store 4 frames of edge feature data. When moderate screen distortion occurs, the 300 MHz frequency provides stronger real-time processing capabilities, and the 1024 kilobyte cache supports more complex feature analysis. When the screen is severely distorted, the 400 MHz frequency combined with the 1.5 volt operating voltage provides peak computing performance, and the 2048 kilobyte cache realizes the correlation analysis of multiple frames of data. The sliding window monitoring mechanism realizes smooth tracking of the degree of screen distortion. The window length of 50 milliseconds contains 6 frames of image data, and the evaluation results are updated every 10 milliseconds. When moderate screen distortion is detected, the power budget is increased from the 10 watt baseline value to 12 watts, and further increased to 15 watts when severe screen distortion occurs. As display quality improves, the power budget gradually decreases at a rate of 0.5 watts every 10 milliseconds, ensuring picture stability while achieving power consumption management.

[0042] S107, obtaining the use environment parameters of the liquid crystal display screen in real time, adaptively adjusting the power distribution mode of the driving circuit and the suppression circuit according to the environment parameters, and increasing the power budget of the suppression circuit in extreme environments.

[0043] The environmental parameter values ​​collected by the temperature sensor, the voltage monitoring unit and the light sensor are obtained, and the environmental parameter vector is obtained through standardization processing; the environmental parameter vector is processed by using the support vector regression method, and the environmental score value is calculated according to the degree to which the temperature value, the voltage value and the brightness value exceed the normal range; the power budget is allocated for the environmental score value, if the environmental score value is greater than the preset threshold, it is allocated to the driving circuit and the suppression circuit according to the first power ratio, if the environmental score value is less than the preset threshold, it is allocated according to the second power ratio; the change trend of the environmental score value is obtained through a sliding time window, if the mean environmental score value in the sliding time window continues to rise and the score increment of the adjacent window is greater than the set threshold, the power budget allocation ratio is adjusted according to the preset step size.

[0044] Exemplarily, the ambient temperature value is collected every 100 milliseconds by a temperature sensor, and the normal temperature range is 0 to 40 degrees Celsius. The voltage fluctuation value is collected every 50 milliseconds from the power monitoring unit, and the normal fluctuation range is 90% to 110% of the nominal voltage. The ambient brightness value is collected every 200 milliseconds by a light sensor, and the normal brightness range is 50 to 1000 lux. The temperature value, voltage value, and brightness value are standardized to the interval of 0 to 1 to construct an environmental parameter vector. Support vector regression is used to process the environmental parameter vector, and an environmental scoring model is established based on the temperature value, voltage value, and brightness value. The environmental parameters within the normal range are scored as 1 point, and the parameters beyond the normal range are deducted according to the degree of excess. For every 5 degrees Celsius exceeding the temperature, 0.2 points are deducted, for every 5% exceeding the voltage, 0.1 points are deducted, and for every 20% exceeding the brightness, 0.1 points are deducted, and an environmental scoring value between 0 and 1 is output. The power budget is dynamically allocated according to the environmental score value. The total power budget is fixed at 20 watts. When the environmental score is 1, the drive circuit is allocated 12 watts and the suppression circuit is allocated 8 watts. The suppression circuit power increases by 1 watt for every 0.1 point decrease in the environmental score. When the environmental score is lower than 0.5 points, the drive circuit is reduced to 8 watts and the suppression circuit is increased to 12 watts to ensure the display quality in harsh environments. A 200-millisecond sliding time window is used to monitor the trend of environmental parameter changes. 20 sets of environmental data are collected in the window. When the average environmental score of 5 consecutive windows continues to rise and the increment of the adjacent window score is greater than 0.05, the suppression circuit power is reduced by 0.5 watts in each sliding cycle until it returns to the standard allocation ratio, realizing adaptive adjustment of the power budget. The environmental adaptability of the LCD display is reflected in the dynamic response to external conditions such as temperature, voltage and light. The 100-millisecond sampling period of the temperature sensor ensures that the ambient temperature changes are captured in a timely manner. In indoor office scenarios, the temperature is usually stable at 25 degrees Celsius, while the temperature may fluctuate to 45 degrees Celsius when used outdoors. Power monitoring pays more attention to instantaneous changes. The sampling interval of 50 milliseconds can detect power grid fluctuations. When the standard 12 volt power supply is used, the normal fluctuation range is between 10.8 and 13.2 volts. The light sensor adapts to the visual adaptation time of the human eye. The sampling period of 200 milliseconds reflects the changes in ambient light. Indoor lighting is usually 300 lux, and direct sunlight can reach 2000 lux. Support vector regression uses radial basis kernel function to construct nonlinear mapping to comprehensively evaluate the severity of the environment. In a standard office environment, the environmental score corresponding to 25 degrees Celsius temperature, 12 volts voltage, and 300 lux lighting is 1 point. When the temperature rises to 35 degrees Celsius, 0.4 points will be deducted. At the same time, if the voltage fluctuates to 11 volts, 0.1 points will be deducted. If the lighting increases to 600 lux, 0.1 points will be deducted, and the comprehensive environmental score will be reduced to 0.4 points. The deduction weight of each parameter reflects the degree of its impact on the display quality. The slow response of the liquid crystal caused by temperature changes is the most significant.The power budget allocation strategy is optimized for different environmental conditions. Under the total power limit of 20 watts, the driving circuit in the standard environment occupies 60%, that is, 12 watts, which is enough to maintain normal display. When the environmental score drops to 0.7 points, the suppression circuit power is increased to 11 watts to enhance the protection of display quality. When the environmental score falls below 0.5 points, the suppression circuit is allocated with a high priority of 12 watts to ensure that basic display functions are maintained under harsh conditions. The power adjustment is in steps of 1 watt to avoid screen jitter caused by frequent switching. The environmental parameter monitoring uses a 200 millisecond sliding window, which includes 2 temperature sampling values, 4 voltage sampling values, and 1 light sampling value. When the air conditioner is turned on and the room temperature gradually decreases, the score of 5 consecutive monitoring windows gradually increases from 0.6 to 0.8, and the score increment of each window exceeds 0.05. At this time, the suppression circuit power is reduced by 0.5 watts every 200 milliseconds, and it is reduced from 11 watts to 9 watts after 4 cycles. This gradual adjustment not only ensures a smooth transition of display quality, but also realizes dynamic optimization of power consumption.

[0045] It will be apparent to those skilled in the art that the present application is not limited to the details of the exemplary embodiments described above, and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims be included in the present application. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for analyzing and suppressing screen noise of a liquid crystal display, characterized in that: The method comprises: Acquire real-time working status information and properties of the liquid crystal display, and determine the current power requirement of the liquid crystal display according to the real-time working status and properties of the liquid crystal display, wherein the real-time working status information includes brightness, contrast and color saturation of the display screen, and the properties include refresh rate and resolution; The support vector machine algorithm is used to process the real-time working status information of the LCD screen, predict the power demand of the driving circuit, and obtain the power demand curve of the driving circuit under different working conditions; Obtain power consumption characteristic parameters of the screen noise suppression circuit, which include the required storage space and the static power consumption of the circuit. A mapping relationship between the power consumption characteristic parameters and the working intensity is established to quantify the power demand under the storage space occupancy rate and the static power consumption conditions. Combining the real-time working status of the LCD screen, the power demand prediction curve of the driving circuit and the power consumption characteristics of the screen flicker suppression circuit, the fuzzy control algorithm is used to design the initial screen flicker suppression method, and the power budget allocation ratio between the driving circuit and the suppression circuit is dynamically adjusted; Based on the designed initial screen flicker suppression method, the screen flicker suppression method is optimized through a neural network algorithm. On the premise of maintaining the suppression effect, the calculation process is simplified and the calculation complexity and power consumption of storage access are optimized to reduce the proportion of the suppression circuit in the entire system power consumption. Determine whether the screen is distorted based on the real-time working status of the LCD screen. When the screen is distorted, dynamically increase the working intensity and power budget of the suppression circuit to eliminate the distorted screen. When the display is normal, reduce the suppression intensity. The operating environment parameters of the LCD display are obtained in real time, the power distribution mode of the driving circuit and the suppression circuit is adaptively adjusted according to the environmental parameters, and the power budget of the suppression circuit is increased in extreme environments.

2. The method according to claim 1, characterized in that The real-time working state information and properties of the liquid crystal display are obtained, and the current power demand of the liquid crystal display is determined according to the real-time working state and properties of the liquid crystal display, wherein the real-time working state information includes brightness, contrast and color saturation of the display screen, and the properties include refresh rate and resolution, including: Receive brightness data of a liquid crystal display screen collected by a multi-point brightness sensor, wherein the brightness data includes temperature values ​​at the four corners and the center of the display screen, a brightness matrix of a current frame of the display screen, a contrast value, and a color saturation value; According to the brightness data, a Butterworth low-pass filter is used to obtain the average value of each parameter of the liquid crystal display screen, and the backlight uniformity value and the frame change rate value are calculated based on the average value; When the temperature value exceeds a preset threshold, the output current of the backlight driving unit is reduced, the driving circuit supply voltage is adjusted according to the frame change rate value, and the signal gain multiple of the color processing unit is compensated according to the color saturation value; A support vector regression method is used to establish a mapping relationship between the backlight uniformity value, the brightness matrix value, the contrast value, the color saturation value and the frame change rate value, so as to obtain a power demand value of the liquid crystal display screen in the current working state; It also includes: collecting panel attribute parameters of the LCD screen when it is working in real time, and comparing the panel attribute parameters with preset standard values ​​to obtain deviation values; using the refresh rate and resolution attributes of the LCD panel as input conditions for power demand calculation, combining the real-time working state deviation value and the panel attribute parameters for linear interpolation calculation, and obtaining the real-time power demand value of the LCD screen in the current working state.

3. The method according to claim 2, characterized in that The panel attribute parameters of the liquid crystal display screen when working in real time are collected, and the panel attribute parameters are compared with preset standard values ​​to obtain a deviation value; the refresh rate and resolution attributes of the liquid crystal panel are used as input conditions for power demand calculation, and linear interpolation calculation is performed in combination with the real-time working state deviation value and the panel attribute parameters to obtain the real-time power demand value of the liquid crystal display screen in the current working state, including: Obtain the brightness value, contrast value and color saturation value collected by the liquid crystal display screen sensor, perform difference calculation based on the values ​​and the reference values ​​in the preset standard parameter table, and obtain the brightness deviation value, contrast deviation value and color saturation deviation value; The refresh rate value and display resolution parameter read by the LCD screen driver chip are normalized to obtain the basic power demand coefficient; Generate a power compensation coefficient using cubic spline interpolation according to the brightness deviation value, contrast deviation value and color saturation deviation value; The power compensation coefficient and the basic power requirement coefficient are combined and calculated through a gradient boosting tree to obtain a power requirement value of the liquid crystal display screen.

4. The method according to claim 1, characterized in that: The support vector machine algorithm is used to process the real-time working state information of the liquid crystal display screen, predict the power demand of the driving circuit, and obtain the power demand curve of the driving circuit under different working states, including: The circuit integrated measurement unit is used to obtain the working state data of the liquid crystal display screen, wherein the working state data includes the brightness value and contrast value output by the display driver chip, the temperature value collected by the temperature sensor, and the voltage value on the current sampling resistor; A support vector machine model is established according to the working state data, and a nonlinear mapping is performed on the working state data through a Gaussian kernel function to obtain a display screen power prediction value; Establishing a reference mapping table for the power prediction value, and using a cubic spline curve to continuously interpolate adjacent measurement points in the reference mapping table to obtain a power demand curve; According to the power value corresponding to the current working state in the power demand curve, the power supply voltage of the driving circuit is adjusted by the bus voltage regulator to obtain the output limit value of the power supply voltage.

5. The method according to claim 1, characterized in that The power consumption characteristic parameters of the screen flicker suppression circuit are obtained, and the power consumption characteristic parameters include the required storage space and the static power consumption of the circuit. A mapping relationship between the power consumption characteristic parameters and the working intensity is established to quantify the storage space occupancy rate and the power demand under the static power consumption conditions, including: Collecting the usage of storage units, power supply voltage values, working current values ​​and temperature values ​​in the screen distortion suppression circuit, and calculating the cache hit rate value based on the collected data; A random forest model is constructed according to the storage space occupancy rate, cache hit rate, temperature value, voltage-current product and the suppression task processing amount, and a power consumption prediction value is obtained through the random forest model; Determine the gear to which the processing capacity value of the suppression task belongs. If the processing capacity value is in the light load interval, the first gear value is obtained. If the processing capacity value is in the medium load interval, the second gear value is obtained. If the processing capacity value is in the heavy load interval, the third gear value is obtained. If the processing capacity value is in the full load interval, the fourth gear value is obtained. The gear value and the power consumption prediction value are used to construct a historical record point, a working gear and power demand mapping function is obtained by fitting the historical record point, and the power demand value corresponding to any workload gear is obtained according to the mapping function.

6. The method according to claim 1, characterized in that The method combines the real-time working state of the liquid crystal display, the power demand prediction curve of the driving circuit, and the power consumption characteristics of the screen flicker suppression circuit, adopts a fuzzy control algorithm to design an initial screen flicker suppression method, and dynamically adjusts the power budget allocation ratio between the driving circuit and the suppression circuit, including: Construct a state evaluation vector based on the screen brightness value, contrast value and color saturation value obtained in real time; Establishing a triangle membership function according to the state evaluation vector, generating a rule mapping table for the membership function, and obtaining a rule reasoning result by a centroid method; The weighted average method is used to calculate the working state parameter weight coefficient corresponding to the rule reasoning result, and the driving circuit power prediction curve is divided into sections according to the weight coefficient; If the product of the weight coefficient and the quality score is greater than a preset threshold, the proportion of the driving circuit power is increased, and the power budget allocation ratio is updated by an exponential sliding average method.

7. The method according to claim 1, characterized in that The initial screen flicker suppression method based on the design is optimized by a neural network algorithm. Under the premise of maintaining the suppression effect, the calculation process is simplified, and the calculation complexity and power consumption of storage access are optimized to reduce the proportion of the suppression circuit in the power consumption of the entire system, including: Acquire the original flower screen image and the suppression result image output by the image data acquisition unit, and record the operation type and data amount of the operation unit according to the original flower screen image and the suppression result image; A convolutional neural network is used to model the operation type and data volume of the operation unit, and a contribution of the operation branch is obtained by calculating the loss function, and a pruning operation is performed on the operation branch whose contribution is less than a threshold of the loss function; For the operation branch after the pruning operation, a local cache is used to store the continuous access data of the operation unit, a cache mapping table is established according to the continuous access data, and a fixed-point conversion is performed on the floating-point operation of the operation unit; The voltage and current values ​​of the operation unit are collected through the power consumption detection circuit, the power consumption curve is calculated according to the voltage and current values, and the image quality score is obtained based on the image edge sharpness and the number of noise points.

8. The method according to claim 1, characterized in that: The method of judging whether a screen distortion phenomenon occurs according to the real-time working state of the liquid crystal display screen, and dynamically increasing the working intensity and power budget of the suppression circuit to eliminate the screen distortion phenomenon when the screen distortion phenomenon is detected, and reducing the suppression intensity when the display is normal, includes: Extracting edge contour feature values ​​according to display screen image data, and calculating edge difference data of adjacent image frames by using the edge contour feature values; Acquire the edge difference data, calculate the noise density value using a sliding window, and construct a feature vector according to the noise density value and the color saturation value; Performing a support vector machine classification operation on the feature vector, and obtaining a screen flower level determination result through the classification operation; Adjust the operating frequency parameter, the computing precision parameter and the storage bandwidth parameter according to the result of the screen distortion level determination. If the edge difference data in the continuous monitoring period is lower than the preset threshold, reduce the operating frequency parameter, the computing precision parameter and the storage bandwidth parameter according to the preset gradual change period. It also includes: real-time monitoring of the display screen of the liquid crystal display screen, and extracting the position and shape characteristics of the edge of the object in the screen by using an image analysis method based on edge detection; By calculating the displacement and deformation degree of edge features between adjacent frames, it is determined whether there is a screen distortion phenomenon such as breakage or ghosting in the picture. If a screen distortion phenomenon is detected, the working intensity parameters of the suppression circuit are dynamically adjusted according to the quantitative index of the screen distortion degree, and its allocation ratio in the overall power budget is increased accordingly until the quantitative index of the screen distortion phenomenon is reduced below the preset threshold.

9. The method according to claim 8, characterized in that The real-time monitoring of the display screen of the liquid crystal display screen uses an edge detection-based image analysis method to extract the position and shape characteristics of the edge of the object in the screen; by calculating the displacement and deformation degree of the edge characteristics between adjacent frames, it is determined whether there is a screen distortion phenomenon such as breakage and ghosting in the screen; if the screen distortion phenomenon is detected, the working intensity parameter of the suppression circuit is dynamically adjusted according to the quantitative index of the screen distortion degree, and its allocation ratio in the overall power budget is increased accordingly, until the quantitative index of the screen distortion phenomenon is reduced to below a preset threshold, including: Acquire image data of a liquid crystal display screen through an image processing unit, extract edge contours using a Laplace operator according to the image data, and acquire feature point coordinate values ​​and curvature values ​​for the edge contours; The displacement vector between consecutive frames is calculated by using the pyramid optical flow method according to the feature points, and if the length of the displacement vector exceeds a preset threshold, it is determined to be a ghosting area, and if the displacement direction deviation is less than a preset angle, it is determined to be the same ghosting group; The ghosting index is obtained by counting the pixel ratio in the ghosting area, and the fracture index is obtained by calculating the number of curvature discontinuity points in the edge contour. The quantitative value of the screen noise degree is obtained according to the ghosting index and the fracture index according to the preset weights; An average value is calculated using a sliding window according to the quantized value. If the average value exceeds a first preset threshold, the power budget of the suppression circuit is increased. If the average value is lower than a second preset threshold, the power budget is reduced to a reference level.

10. The method according to claim 1, characterized in that The real-time acquisition of the use environment parameters of the liquid crystal display screen, adaptively adjusting the power allocation mode of the driving circuit and the suppression circuit according to the environment parameters, and increasing the power budget of the suppression circuit in extreme environments, includes: Obtain environmental parameter values ​​collected by the temperature sensor, the voltage monitoring unit, and the light sensor, and obtain an environmental parameter vector through standardization; The environmental parameter vector is processed by using a support vector regression method, and an environmental score value is calculated according to the degree to which the temperature value, the voltage value and the brightness value exceed the normal range; Performing power budget allocation according to the environmental score value, if the environmental score value is greater than a preset threshold, allocating the power budget to the driving circuit and the suppression circuit according to a first power ratio, and if the environmental score value is less than the preset threshold, allocating the power budget according to a second power ratio; The changing trend of the environmental score value is obtained through a sliding time window. If the mean environmental score in the sliding time window continues to rise and the increment of the score in an adjacent window is greater than a set threshold, the power budget allocation ratio is adjusted according to a preset step size.

Images