Display resource distribution system based on fuzzy logic

Abstract

The invention belongs to the field of computer system resource allocation based on a specific calculation model, and discloses a display resource allocation system based on fuzzy logic, which comprises the following steps that: the system acquires an acceleration vector of user interaction, and performs authenticity discrimination based on direction consistency of the acceleration vector to generate weighted interaction acceleration; the fuzzy logic reasoning engine maps the weighted acceleration into a resource allocation priority, and meanwhile, the operation cycle of the engine is dynamically adjusted by the core temperature of the system so as to compensate calculation delay; according to the method, resource scheduling is changed from an application state to a user physical intention subjected to authenticity verification, deep synchronization of system response and user attention is realized, delay and lagging of key interaction are avoided, and smooth and harmonious qualitative change of man-machine interaction experience is achieved.

Description

Technical Field

[0001] The present invention relates to a display resource allocation system based on fuzzy logic, belonging to the technical field of computer system resources based on specific calculation models. Background Art

[0002] In current mainstream display systems such as electronic whiteboards, collaborative interactive large screens, etc., the allocation of internal computing resources, such as CPU time or GPU rendering permissions, usually follows a mature and widely accepted operating logic, that is, the priority of its resource acquisition is determined by the system status presented by the application itself. This logic uses whether the application is in the foreground and whether it is performing high-load data processing or content rendering, and other states that can be directly identified by the operating system, as the core basis for judging the urgency of resource needs. This approach reflects the simplicity and universal applicability of its design when processing traditional computing tasks with clear boundaries.

[0003] However, as interaction modes evolve toward high dynamics, high density, and multi-tasking, the inherent limitations of the aforementioned resource allocation logic begin to emerge, especially in scenarios such as remote teaching design reviews that require frequent and precise user operations. In a typical teaching interaction scenario, a teacher may be playing a high-definition video that takes up most of the screen as a background while using a brush tool to quickly and forcefully circle and annotate the video to emphasize key content. In this case, based on existing logic, the system will determine that the video playback application is a high-load foreground task and devote the majority of resources to it. The brush tool, however, is identified as a low-load operation due to its small rendering area and small pixel changes. The system's resource allocation results in a fundamental deviation from the user's true intention focus, resulting in delayed brush response and broken handwriting, seriously affecting the fluency of the core teaching task. This phenomenon reveals a profound contradiction in the existing technical paradigm: it equates the application's computational characteristics with the user's intention priority, but ignores the unencoded dynamic information contained in the user's physical interaction behavior that most directly reflects the intensity and urgency of their intention.

[0004] In the face of such problems, an intuitive improvement idea is to indiscriminately increase the response priority of all user input events, but this approach does not touch the essence of the problem. It cannot effectively distinguish between a leisurely drag and a quick writing with urgent intentions. When multiple points or multiple applications interact concurrently, it may cause new resource competition chaos due to frequent and indiscriminate priority increases, making the overall operation of the system more disordered. Therefore, how to establish a resource allocation mechanism that can transcend the limitations of the application layer state, so that it can directly perceive and understand the dynamic characteristics of the user's physical interaction behavior, and convert this understanding into a fine-grained resource scheduling strategy that is synchronized in real time with the user's actual intention intensity, thereby solving the problem of mismatch between system response logic and user attention focus, has become the technical problem to be solved by the present invention. Summary of the Invention

[0005] The present invention provides a display resource allocation system based on fuzzy logic, the main purpose of which is to solve the problem that the existing technology relies on application status for resource allocation, thereby causing the system response to be inconsistent with the user's actual intention focus.

[0006] To achieve the above objectives, the present invention provides a display resource allocation system based on fuzzy logic, comprising: an interactive dynamic feature acquisition module configured to acquire a coordinate time series generated by user interaction at a sampling frequency of not less than 120 Hz, and calculate an acceleration vector sequence based on the coordinate time series; an intention authenticity identification unit configured to calculate a quantitative index representing the directional consistency of the acceleration vector sequence based on the acceleration vector sequence output by the interaction dynamic feature acquisition module within the time window, and to weight the acceleration amplitude of the interaction behavior according to the quantitative index to generate a weighted interaction acceleration; a fuzzy logic inference engine module configured to take the weighted interaction acceleration output by the intention authenticity discrimination unit as a first fuzzification input and, according to a fuzzy rule base stored in a memory, map the first fuzzification input into a resource allocation priority value during its operation cycle; A temperature sensing module, used to obtain a real-time temperature value of the system core processor; a cycle adaptive adjustment unit configured to dynamically adjust the operating cycle of the fuzzy logic inference engine module according to the real-time temperature value and a temperature and calculation delay characteristic model stored in the memory; A resource dynamic scheduling module is configured to allocate the system's computing resources according to the resource allocation priority value output by the fuzzy logic reasoning engine module.

[0007] Preferably, the intention authenticity identification unit calculates a quantitative index of directional consistency, specifically: calculating the cosine value of the directional angle between each pair of consecutive acceleration vectors in the acceleration vector sequence within the time window; and performing arithmetic averaging on all calculated cosine values ​​to obtain an average cosine value as a quantitative index of directional consistency.

[0008] Preferably, the intention authenticity identification unit generates the weighted interaction acceleration through the following relationship: , ,in, is the original magnitude of the acceleration of the interaction behavior, is the quantitative index of directional consistency, namely the average cosine value. A data file stored in memory and A monotonically increasing weight function whose value range is from zero to one.

[0009] Preferably, the cycle adaptive adjustment unit dynamically adjusts the operating cycle of the fuzzy logic reasoning engine module, specifically: determining an expected calculation delay based on the real-time temperature value and the temperature and calculation delay characteristic model; and subtracting the expected calculation delay from a baseline operating cycle to obtain a new, shortened operating cycle.

[0010] Preferably, the fuzzy logic reasoning engine module is further configured to: take a numerical value representing the content rendering complexity associated with the interactive behavior as a second fuzzification input, and combine the weighted interaction acceleration and the content rendering complexity values ​​to jointly determine the resource allocation priority value.

[0011] Preferably, the fuzzy rule base includes the following rule: when the fuzzified value of the weighted interaction acceleration falls into a high-intensity interaction interval, the resource allocation priority value is determined to be a highest priority value, and the execution of this rule is independent of the state of any other fuzzified input.

[0012] Preferably, the system also includes a collaborative intention aggregation unit, which is configured to: confirm that a collaborative event occurs when the interactive dynamic feature acquisition module detects at least two interactive behaviors with different spatial positions within a set time window, and the intention authenticity identification unit outputs high-intensity weighted interactive acceleration for at least two interactive behaviors; and based on the confirmation of the collaborative event, trigger a collaborative allocation strategy that concentrates computing resources in the area where at least two interactive behaviors are located.

[0013] Preferably, the system also includes a cognitive load state tracking unit, which is configured to: calculate an exponential moving average representing the user's continuous cognitive load based on a series of resource allocation priority values ​​output by the fuzzy logic reasoning engine module over a period of time; and when the exponential moving average exceeds a load threshold stored in the memory, automatically adjust the presentation mode of all notification information in the system except alarm information to reduce cognitive interference to the user.

[0014] Preferably, the interactive dynamic feature acquisition module is also configured to: during the period when no user interaction is detected, set the acceleration of the interactive behavior to a zero value, so that the weighted interactive acceleration output by the intention authenticity identification unit is also zero, so that the decision basis of the fuzzy logic reasoning engine module is switched to other non-interaction-related fuzzified inputs.

[0015] Preferably, the resource dynamic scheduling module is configured to: implement the allocation of computing resources by adjusting the CPU time slice allocation and the GPU rendering queue priority of the application process corresponding to the interactive behavior.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention introduces the physical acceleration generated when the user interacts with the display interface as a continuous and dynamic physical feature into the resource allocation decision-making process, and establishes a direct correlation between this physical dynamic feature and the resource allocation priority within the system through a fuzzy logic reasoning mechanism. This approach transforms the core basis for system resource scheduling from relying on the relatively static or slowly changing operating state of the application itself to directly responding to the user's instantaneous intention intensity during physical operations. When the user interacts at high speed and forcefully, the system no longer ignores its resource request because the application to which the interaction belongs is in the background or appears to be underloaded. This avoids delays or freezes in key operations caused by the mismatch between system cognition and user intention, and keeps the system's response logic and the user's focus of attention synchronized in time.

[0017] 2. Based on the acceleration vector of interactive behavior as a measure of intention intensity, the present invention further introduces the consideration of the directional stability of the acceleration vector within a continuous time window. By coupling the magnitude characteristics of the acceleration with the continuity characteristics of its direction, the system obtains the ability to identify the authenticity of intentions. It can distinguish between fast strokes with stable direction driven by the user's clear intention and non-intentional jitter with scattered direction caused by physiological or environmental factors. In this way, the system avoids misinterpreting high-frequency jitter pseudo-signals as high-intensity user intentions, thereby suppressing unnecessary resource priority increases and system load fluctuations, making its perception of the user's true intentions more adaptable and reliable under different users and changing environments.

[0018] 3. The fuzzy logic reasoning engine constructed by the present invention continuously generates a resource allocation priority output sequence during operation, which is further utilized as a new information source. By analyzing the statistical characteristics of the priority sequence over a period of time, the system can indirectly measure the persistence level of the user's current cognitive load. When it is recognized that the user is in a high-intensity interaction state, that is, a high cognitive load state, for a long time, the system will actively adjust the presentation method of non-critical notifications to reduce cognitive interference. This design will be used for the control output of micro-operation responses and reused as perceptual input for understanding the macro-user state. Without adding any additional hardware or perception modules, the system can be extended from ensuring the smoothness of the operation level to protecting the user's immersive work flow, and the human-computer interaction experience will therefore be more coherent and harmonious. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a diagram of the integrated architecture of system perception, decision-making and adaptive regulation of the present invention; Figure 2 A flowchart of mapping user physical intent to system resource priority according to the present invention; Figure 3 This is a two-stage adaptation procedure diagram for the system model of the present invention, facing hardware and users. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. It should be noted that the embodiments in this application and the features in the embodiments can be combined with each other unless there is a conflict.

[0021] The present invention provides a fuzzy logic-based display resource allocation system. Its overall architecture is based on a specific computational model and aims to directly map the dynamic features contained in user physical interaction behaviors into resource allocation decisions within the system. The system primarily comprises an interactive dynamic feature acquisition module, an intent authenticity verification unit, a fuzzy logic inference engine module, a temperature sensing module, a cycle adaptive adjustment unit, and a resource dynamic scheduling module. Under specific configurations, the system may also integrate a collaborative intent aggregation unit and a cognitive load state tracking unit. The top-level data flow begins with the interactive dynamic feature acquisition module's perception of user physical actions. The generated acceleration vector sequence, after weighted adjustment by the intent authenticity verification unit, is fed as the core input into the fuzzy logic inference engine module. The engine, combined with auxiliary inputs such as content rendering complexity, generates resource allocation priority values ​​based on a built-in fuzzy rule base. Simultaneously, the engine's own operating cycle is dynamically regulated by a compensation loop formed by the temperature sensing module and the cycle adaptive adjustment unit. Ultimately, the resource dynamic scheduling module allocates resources of the central processing unit and graphics processing unit in real time based on these priority values.

[0022] In a typical remote teaching or collaborative design scenario with high requirements for interactive fluency, the existing resource allocation method often leads to a mismatch between system response and the user's actual intention focus due to its dependence on application layer status. To address this technical bottleneck, the interactive dynamic feature acquisition module of the present invention is configured to continuously acquire the coordinate time series generated when the user interacts with the display interface through media such as a stylus or finger at a sampling frequency of not less than 120 Hz. This frequency ensures that even the details of fast strokes can be captured without omission. In order to convert this raw coordinate data into a physical quantity that can represent the intensity of the user's intention, the module calculates an acceleration vector sequence in real time by performing a second-order difference operation on continuous coordinate points. For example, the acceleration vector at any moment Available through The calculation results are as follows: Represents the coordinate vector of the current moment, The sampling time interval is used as the sampling time interval. In this way, the system converts the user interaction force that was originally imperceptible into a measurable and analyzable acceleration physical feature, building an input foundation based on physical reality for subsequent intention understanding.

[0023] In a real-world usage environment, the user's physiological hand tremors or vibrations caused by environmental factors may generate high-amplitude acceleration pseudo-signals. If not discriminated, the system will mistakenly interpret these meaningless tremors as high-intensity interaction intentions. To address this challenge, the solution introduces an intent authenticity discrimination unit. The core mechanism of this unit is to draw on the principles used in the field of signal processing to measure the stability of time series to quantitatively judge the authenticity of interaction intentions. The specific procedure is to first, in a system containing For example, Within the sliding time window of the latest acceleration vector, calculate each pair of consecutive acceleration vectors and The cosine value of the angle between the directions, then all the calculated The cosine values ​​are arithmetic averaged to obtain a quantitative index that can characterize the directional consistency of the acceleration vector sequence in a short period of time, namely the average cosine value. , in view of this, the unit then uses a preset weight function The original acceleration amplitude Adjust to produce the final weighted interaction acceleration , and the relationship is , this weight function Designed as a A monotonically increasing function whose range is between zero and one and is preset in the system memory in the form of a lookup table or analytical expression. The value of is close to 1, indicating that the user's stroke direction is stable and the intention is clear. The value of also approaches 1, thus completely retaining the original acceleration amplitude. On the contrary, when When the value of is low, it indicates that the interaction direction is confused and may be unintentional jitter. The value of approaches 0, thereby effectively suppressing the intensity of the acceleration false signal. Through this dynamic weight adjustment based on the directional stability of the acceleration vector, the system acquires the ability to identify the authenticity of the user's intention and provides a purified high-confidence input signal for subsequent fuzzy logic reasoning.

[0024] After obtaining the weighted interaction acceleration that reflects the strength of user intention Finally, how to convert it into resource allocation instructions that the operating system can understand and execute is the core of this computing model. Given that the mapping relationship from physical interaction intensity to resource priority requirements is fuzzy and nonlinear, the system uses a fuzzy logic reasoning engine module to undertake this task. The fuzzification stage of the engine will be the weighted interaction acceleration output by the intention authenticity identification unit. As the first fuzzy input, and its numerical value is mapped to a fuzzy language set such as smooth, agile, and rapid according to the membership function. In order to achieve more refined decision-making, the engine can also be configured to use a numerical value representing the complexity of content rendering associated with interactive behavior as the second fuzzy input. This numerical value can be obtained from the graphics processor driver interface and is also mapped to fuzzy sets such as light, medium, and heavy. On this basis, the fuzzy logic reasoning engine module performs reasoning based on a fuzzy rule base preset in the memory. The rule base consists of a series of IF-THEN rules, for example, IF is urgent THEN priority is overwhelming. This rule is set to have a high execution weight. That is, when the system perceives that the user is performing a high-intensity and urgent interaction, it will directly determine the resource allocation priority as a highest priority value. The execution of this rule is independent of the state of any other fuzzy input. Other rules such as IF is agile AND is weight THEN priority is high, which covers all other working conditions. After the reasoning and defuzzification process, the engine finally outputs a resource allocation priority value with a clear range. This value provides a direct quantitative basis for subsequent resource scheduling.

[0025] When electronic equipment operates under high load for a long time, the temperature increase of its core processor may trigger the thermal throttling mechanism, resulting in increased computing delay. In order to compensate for the impact of this physical environment change, the system integrates an adaptive adjustment mechanism consisting of a temperature sensing module and a periodic adaptive adjustment unit. The temperature sensing module is responsible for obtaining a real-time core processor temperature value using the built-in sensor of the processor, while the periodic adaptive adjustment unit dynamically adjusts the operating cycle of the fuzzy logic inference engine module based on the real-time temperature value and a temperature and computing delay characteristic model preset in the memory. The characteristic model is established through an offline calibration experiment with controllable ambient temperature: at different temperature steps, the fuzzy logic inference engine module is driven to continuously execute thousands of standard inference operations and accurately measure its average computing delay, thereby obtaining a set of core processor temperatures related to the specific hardware. Compute average latency By fitting a polynomial to the discrete data points, a function that can describe the relationship between the two is finally generated. And build a lookup table, solidified in the system memory; based on this, the unit uses a back-compensation deterministic procedure to set the next operating cycle of the fuzzy logic inference engine , which is calculated from a baseline operating cycle , for example, 20 milliseconds minus the expected computational delay ,Right now For example, when the core processor temperature is 40°C, the expected delay calculated by the model is is 0.5 milliseconds, the new operating cycle is , when the core processor temperature rises to 85°C due to high load, the expected delay calculated by the model If the cycle time increases to 3.0 milliseconds, the system will automatically shorten the operation cycle to In this way, when the processor temperature rises and the expected delay increases, the system will actively shorten the decision-making interval and use the increase in decision-making frequency to offset the decrease in single decision-making speed, thereby maintaining the total external response time at a stable level.

[0026] After the fuzzy logic inference engine module outputs the final resource allocation priority value, the resource dynamic scheduling module is responsible for converting this decision into the actual allocation of system computing resources. This module interacts with the kernel scheduler of the operating system and implements resource allocation by adjusting one or both of the CPU time slice allocation and the GPU rendering queue priority of the application process corresponding to the current interactive behavior according to the received priority value. For example, a brush application process that is assigned a high priority value, such as 95, can temporarily adjust its nice value in the Linux-like system to -20, and its related rendering instructions can be temporarily adjusted to -20. It is placed at the front end of the graphics processor task queue, thereby obtaining more processor resources to ensure the smoothness and instantaneous response of its handwriting rendering; in order to cope with the scenario of multi-user collaborative work, the system can also include a collaborative intention aggregation unit. When the interactive dynamic feature acquisition module detects at least two interactive behaviors with different spatial positions within a set time window, and the intention authenticity identification unit outputs high-intensity weighted interactive acceleration for at least two interactive behaviors, the aggregation unit confirms that a collaborative event has occurred. Based on this confirmation, the system will trigger a collaborative allocation strategy, for example, aggregating computing resources to the area where at least two interactive behaviors are located. domain to ensure the overall smoothness of the entire collaborative task; in addition, to protect the user's cognitive state, the system can also integrate a cognitive load state tracking unit, which reuses the output of the fuzzy logic reasoning engine and takes a series of resource allocation priority values ​​output within a period of time, such as 30 seconds, as input signals. It calculates an exponential moving average of the sequence to represent the user's continuous cognitive load. When this moving average exceeds a preset load threshold stored in the memory, it indicates that the user is in a high-intensity focused work state. The system will automatically adjust the presentation of all notification information except key alarm information, such as pop-up windows. Notifications are downgraded to static icons in the status bar, thereby reducing cognitive interference to users without interrupting information transmission. During the period when no user interaction is detected, the interactive dynamic feature acquisition module is configured to set the acceleration of the interactive behavior to a zero value, which makes the weighted interaction acceleration output by the intent authenticity identification unit also zero. In this non-interactive state, the decision-making basis of the fuzzy logic reasoning engine will smoothly switch to other non-interaction-related fuzzy inputs, such as relying solely on content rendering complexity for decision-making, thereby falling back to a resource allocation logic that is closer to the traditional method, ensuring the consistency of the system's behavior under different interaction modes.

[0027] Example 1: In a remote teaching application scenario, a teacher is demonstrating to students a high-definition physics experiment video that occupies most of the screen. The video content is the motion trajectory of particles in a cloud chamber, and its decoding and rendering processes continue to occupy system computing resources. When the video plays to a critical moment, the teacher needs to use a stylus to make a small but fast and powerful circle on the video screen to highlight a fleeting particle trajectory. In a system that adopts a traditional resource allocation method, the brush tool will be judged as a low-load operation because the rendering area has a small pixel change, and thus insufficient computing resources will not be allocated to it, resulting in delays and freezes in the teacher's handwriting, and the transmission process of the teaching intention is significantly disturbed. When the system using the technical solution of the present invention faces this working condition, its interactive dynamic feature acquisition module captures a series of coordinate time series generated by the teacher's rapid circle and converts it into an acceleration vector sequence. The intention authenticity identification unit then processes the vector sequence and obtains a high directional consistency quantitative index close to 1 by calculating the average cosine value of the directional angle between consecutive vectors. This indicator is used to characterize that the high acceleration signal comes from a stroke with a stable direction and clear intention. Accordingly, the weight function Output a weight value that is also close to 1, so that the original acceleration amplitude of the brush interaction is passed on almost losslessly, generating a high-intensity weighted interaction acceleration ; This weighted interaction acceleration and the numerical value representing the high content rendering complexity of video playback are simultaneously input into the fuzzy logic reasoning engine module.

[0028] Within this module, a rule with a high execution weight from the preset fuzzy rule library is triggered. Specifically, when the fuzzified value of the weighted interaction acceleration falls within a high-intensity interaction interval, the resource allocation priority value is assigned the highest priority. The execution of this rule shifts the core basis for system resource allocation from rendering area at the application level to the intensity of user intent at the physical interaction level, effectively managing resource competition between high-load background applications and high-intent foreground operations. Based on this rule, the engine outputs a highest-priority value. Upon receiving this value, the dynamic resource scheduling module immediately redirects computing resources from the background video playback process to the interaction process associated with the brush tool. Ultimately, each quick circle made by the teacher is instantly and seamlessly displayed on the screen, synchronized with the physical movement of the pen tip, clearly capturing and presenting the fleeting particle trajectories. The system's computing resources are synchronously converged toward the user's focus the moment it shifts, ensuring consistency between the human-computer interaction process and the cognitive flow of the teaching task.

[0029] Example 2: In order to objectively verify the improvement effect of the technical solution of the present invention on the response delay of key interactive events in the scenario where high load and high intensity interaction coexist, this comparative experiment is set up. The purpose of the experiment is to quantitatively compare the rendering delay of high-intention intensity handwriting interaction in the simulated remote teaching scenario between the experimental group using the resource allocation system of the present invention and the control group using the standard operating system resource allocation method; the experimental platform is based on a computer equipped with a multi-touch display screen, whose core processor is a standard quad-core configuration and the operating system is a general Linux distribution. In order to ensure the consistency and reproducibility of the interactive input, a six-axis robotic arm is used to replace the human hand to perform the stylus operation, and the stylus clamped at the end is set to The peak acceleration and The path direction stability is tested by drawing a 10-cm straight line in the specified area of ​​the screen to simulate an eager writing action. The setting of this acceleration value is based on the effective simulation of high-intensity user interaction. This value corresponds to the typical value of the high-intensity interaction interval obtained from the statistical analysis of user interaction behavior. A high-speed camera with a frame rate of 240 Hz is facing the screen to capture the process from the physical contact of the stylus to the first rendering of the corresponding handwriting on the screen, and calculates the handwriting rendering delay through frame-by-frame analysis with a time accuracy of 4.17 milliseconds. In the experiment, a high-load background task, a 60 Hz 4K video, was set to play in a loop in 80% of the screen area to continuously occupy system computing resources.

[0030] The experimental process first conducted a benchmark test without background load in the two groups, and recorded that the handwriting rendering delay of both was in the range of 8 milliseconds to 13 milliseconds; then, the background task of 4K video playback was started in both groups, and the test was carried out after the system load stabilized. It was recorded that the average handwriting rendering delay of the control group increased to 79.23 milliseconds, while under the same load conditions, the average handwriting rendering delay of the experimental group was only 12.51 milliseconds; the generation of this difference is directly related to the internal operating status of the experimental group system. In the test of the experimental group, its fuzzy logic inference engine module detected the high-intensity weighted interaction acceleration generated by the brush interaction. , its peak value reaches When the brush process is activated, the resource allocation priority value of the brush process is instantly increased to 96 according to the internal rules, thereby triggering the resource dynamic scheduling module to intervene. To further verify the effectiveness of the intention authenticity identification unit, an interference test was added to the experimental group, that is, while the robotic arm was drawing a line, a slight vibration was applied through the vibration device. Under this working condition, the rendering delay of the experimental group for the intentional brush interaction remained at a low level of 16.68 milliseconds. The reason is that the intention authenticity identification unit weights the acceleration signal generated by the vibration and unstable direction. Substantially lowered so that its corresponding weighted interaction acceleration Suppressed in level, thereby avoiding incorrect resource scheduling; experimental data show that the technical solution of the present invention can, in a computing environment with high-load background tasks, perceive the physical dynamic characteristics of user interaction behavior and identify the authenticity of intention, and then use the fuzzy logic reasoning model to dynamically allocate system resources, thereby controlling the response delay of key interactive operations at a continuously low level.

[0031] Example 3: This example combines Figures 1 to 3 , a display resource allocation system based on fuzzy logic is described, such as Figure 1 As shown, the data flow of the system starts with the collection of the user's physical interaction coordinate time series, and is processed by the interactive dynamic feature acquisition module. The module collects interaction points at a frequency of not less than 120Hz and calculates the acceleration vector sequence; this sequence is then passed to the intention authenticity identification unit, which generates weighted interaction acceleration based on the directional consistency of acceleration to distinguish intentional operations from unintentional jitter. The weighted acceleration signal serves as the core input and is combined with another second fuzzy input representing the complexity of content rendering to enter the fuzzy logic reasoning engine module. The engine serves as the core decision-making unit and maps the input signal to the final resource. Assigning priority values. The operation of the inference engine is also regulated by a compensation loop. The temperature perception module is responsible for obtaining the real-time temperature value of the system's core processor and providing this data to the cycle adaptive adjustment unit. The latter dynamically adjusts the engine's operating cycle according to the temperature and delay model to compensate for the calculation delay. Finally, the resource dynamic scheduling module makes actual adjustments to the CPU / GPU system resources based on the priority value output by the engine. In addition, the system also includes a cognitive load state tracking unit, which calculates the user's cognitive load based on the priority sequence over a period of time, and adjusts non-critical notifications accordingly to reduce cognitive interference.

[0032] like Figure 2As shown in the figure, the interactive dynamic feature acquisition module calculates the acceleration vector by high-frequency sampling of touch coordinates, generates acceleration data, and outputs it to the intent authenticity identification unit. This unit distinguishes true intent from jitter by analyzing directional consistency, and then generates a weighted signal. This signal, along with external information about content rendering complexity, is fed into the fuzzy logic inference engine module as input. This module is the core of decision-making and outputs a quantized priority value. Simultaneously, an independent temperature monitoring loop operates continuously. The temperature sensing module monitors the processor temperature and generates temperature data, which is then passed to the cycle adaptive adjustment unit for delay compensation and dynamic cycle parameters to adjust the inference engine's operating rhythm to ensure stable response. The priority output by the inference engine is obtained by the resource dynamic scheduling module. This module allocates processor time slices and adjusts rendering queue priorities to produce the system's final output: an optimized resource allocation plan. The figure also shows two scalable units in dashed boxes: the collaborative intent aggregation unit for multi-point interaction detection and collaborative resource allocation, and the cognitive load status tracking unit for monitoring user load and performing intelligent notification management. Both units interact with the fuzzy logic inference engine module.

[0033] like Figure 3 As shown, first, it is necessary to build a temperature-computation delay characteristic model, that is, in a temperature-controlled environment, measure and fit the inference calculation delay at different core temperatures, and generate a lookup table or function. Secondly, it is necessary to calibrate the weighted interaction acceleration. Fuzzy membership function, that is, simulating interactions of different intensities through the robotic arm and performing probability density analysis on the collected acceleration data to determine the boundaries of fuzzy sets such as smooth, agile, and rapid; at the same time, it is also necessary to calibrate the direction consistency weight function That is, based on standard line drawing and simulated jitter movements, the mapping relationship between the directional consistency index and the weight value is determined. Finally, through regression analysis of the relationship between multiple GPU performance indicators and the overall utilization rate, the optimal weight coefficient is solved to construct a content rendering complexity calculation model. All models formed in this stage will be solidified in the system memory and deployed with the device. When the user device is started for the first time, it enters the first-time self-calibration phase. The system will guide the user to perform high / low intensity interaction tasks and record the priority sequence in this process. It then calculates the priority mean and sets the load threshold to adapt the threshold to the user's personal usage habits. Subsequently, the system automatically performs internal function self-test and timing calibration, and verifies the logical consistency of the fuzzy rule base by injecting virtual touch events and measures the internal data round-trip time. After the self-test passes, the system initialization is completed and officially enters the normal operation mode.

[0034] Example 4: In a specific engineering practice, in order to ensure that the internal computing model of the technical solution of the present invention can achieve the preset operating effect after being applied to a new touch display device with a specific hardware configuration, it is necessary to perform a one-time offline calibration of several models and parameters in the system. This calibration process aims to match the general algorithm logic with the physical and performance characteristics of the specific hardware platform; the calibration process first starts with establishing a temperature and computing delay characteristic model. The construction of this model aims to compensate for the fluctuation of the fuzzy logic inference engine computing time caused by the temperature change of the processor. The experiment is carried out in a shielded room with controllable ambient temperature. The touch display device to be calibrated is placed in it. The internal temperature sensing module and an external high-precision timer are synchronously connected to a data acquisition host. The calibration procedure is as follows: First, the temperature of the shielded room is set to 25°C. After the temperature of the core processor of the device stabilizes, a test script is used to drive the fuzzy logic inference engine module to continuously perform 1000 standard inference operations. The total time required to complete these 1000 operations is recorded by the timer, and the average computing delay of a single inference is calculated. Then, the temperature of the shielded room is gradually increased in steps of 5°C until the temperature of the core processor of the device reaches its safe operating upper limit, which is 85°C in this embodiment. The above calculation delay measurement process is repeated at each temperature step point. In this way, a set of core processor temperatures related to the specific hardware are obtained. Compute average latency By fitting a polynomial on the discrete data pairs of the data points, a function that characterizes the relationship between the two is generated. Based on this function, a lookup table is constructed and stored in the system memory as the temperature and computational delay characteristic model of the device.

[0035] Next, the core input of the fuzzy logic inference engine module, the weighted interaction acceleration The fuzzy membership function of the fuzzy model is calibrated with the help of a six-axis robotic arm test platform that can perform line drawing at a preset acceleration. The purpose is to make the boundaries of the three fuzzy language sets of smooth, agile, and rapid correspond to the statistical distribution of the interactive behavior. The calibration procedure is as follows: the robotic arm is controlled to cover the range from arrive Within the range of A series of standard line drawing actions are performed on the touch screen with multiple different peak accelerations for the step length. The interactive dynamic feature acquisition module records the original acceleration value corresponding to each action. By performing probability density analysis on a large number of collected acceleration data samples, it can be observed that the data presents multiple clustering areas on the numerical axis. The right boundary of the first main clustering area is defined as the upper limit of the smooth fuzzy set, and the left boundary of the last main clustering area is defined as the lower limit of the agile fuzzy set. The area between the two boundaries belongs to the agile fuzzy set, thereby determining the inflection point of the membership function; at the same time, the weighted interactive acceleration function is used to generate the weighted interactive acceleration function. The weight function Calibration is performed to determine the quantitative index of directional consistency The mapping relationship between the weight value and the intentional stroke is used to distinguish the unintentional jitter. The calibration procedure is as follows: First, the robot arm is controlled to perform a set of directional consistency quantitative indicators. The value is The standard line action, and in this state The output value is set to Secondly, by superimposing a high-frequency random disturbance signal in the motion control program of the robot arm, it can perform a series of directional consistency quantitative indicators. exist to The simulated shaking action fluctuates between The output value is set to Based on these two endpoints and setting the function to be monotonically increasing within the interval, a weight function based on the S-shaped curve can be determined. Through the above calibration procedures, several key models and parameters in the original technical solution are transformed from a general logical framework into an engineering example corresponding to the specific hardware platform.

[0036] Example 5: After being deployed on a specific device, the technical solution of the present invention executes a set of standardized initial self-calibration procedures to adapt its internal model parameters to the user's personal usage habits and the current hardware status. When the device is started for the first time, the system guides the user into an interactive calibration link, which requires the user to perform two preset interactive tasks respectively. One is a low-intensity task of continuously dragging icons at a gentle speed, and the other is a focused high-intensity task of quickly and continuously clicking on multiple targets within a limited time. During this process, the cognitive load state tracking unit continuously records the resource allocation priority value sequence output by the fuzzy logic reasoning engine module, and calculates the priority mean of the low-intensity task stage respectively. and the mean priority of the high-intensity task phase , then, the load threshold in the system is automatically set to a predetermined quantile between these two means, calculated as , where the coefficient It is a preset constant with a value between 0 and 1. In this embodiment, its value is 0.75.

[0037] After this calibration phase, the system further performs an internal function self-test and timing calibration. This process does not require user participation. The system generates a simulated high-intensity weighted interactive acceleration internally. The system directly injects the virtual touch event into the fuzzy logic inference engine module, and then verifies whether the output of the engine is a value in the preset highest priority range to check the logical consistency of the core fuzzy rule base. While doing this verification, the system uses a high-precision internal clock to measure the complete internal data round-trip time from the injection of the virtual event to the receipt of the corresponding priority value by the resource dynamic scheduling module. , the measured time It is used to set the time window threshold for determining whether multiple interaction behaviors occur simultaneously in the collaboration intention aggregation unit, that is, the window is set to .

[0038] Example 6: In a specific engineering implementation, in order to enable the content rendering complexity input used by the fuzzy logic reasoning engine module to quantitatively reflect the actual computing load of different applications on a specific hardware platform, it is necessary to construct a computing model that integrates multiple graphics processor performance indicators into a single value. The construction of this model is framed as an offline optimization procedure with the goal of minimizing prediction error. Its purpose is to solve a set of optimal weight coefficients so that the constructed computing model can accurately predict the actual graphics processor usage; the procedure first selects a set of key performance indicators that can characterize the graphics processor load. In this embodiment, the number of vertices processed per frame is selected. Number of fragment shader instructions executed per frame and the number of texture memory access bytes per frame As basic input variables, in order to eliminate the impact of different dimensions and numerical ranges of each variable, each variable needs to be normalized. The process is to run a set of benchmark programs covering various rendering loads on the target hardware platform and record the maximum value of each variable during the entire test process. , , ,Then, in the actual operation of the system, the normalized index value at any time,is calculated as follows: , , .

[0039] Final content rendering complexity value , obtained by weighted summation of the above normalized indicators, the calculation relationship is: , where the weight coefficient The determination of the ,follows a calibration procedure based on multiple linear regression, in which a series of ,test applications with different rendering loads are run while simultaneously recording ,the basic input variables. The instantaneous value and the overall GPU usage provided by the operating system or driver interface By performing regression analysis on a large number of collected data points, a set of The actual measured value The weight coefficient that minimizes the root mean square error between the values , and satisfies The constraints of this set of calibrated weight coefficients are solidified in the system, thus completing the construction of the content rendering complexity calculation model.

[0040] Example 7: In order to ensure that the allocation instructions of the resource dynamic scheduling module can produce a certain execution effect under different operating system kernel scheduling mechanisms, the module also includes a closed-loop feedback calibration procedure for the scheduling execution effect. This procedure first establishes the resource allocation priority value in the offline calibration stage. A mapping table between a set of kernel-level thread scheduling interface call parameters provided by the operating system , in this relationship, Represents the mapping relationship table, usually implemented in the form of a lookup table or function; Assign the resource priority value output by the fuzzy logic inference engine as the input of the mapping; the mapping relationship converts the input The value is converted into a specific operating system kernel-level thread scheduling interface call and a set of corresponding parameters required to execute the call Among them, the calling parameters include thread priority, CPU core affinity setting and graphics processor context priority; in the real-time operation stage of the system, for each interactive thread that is set to a high priority, the program measures the actual time it takes to complete the rendering or calculation task. and the target response time calculated based on the current interaction acceleration , for example, for an interactive sampling rate of 120Hz, was set to 8.33 milliseconds and compared to calculate the response error , and using a proportional-integral controller, i.e. , to dynamically adjust the priority value output to the mapping table next time, here is the priority adjustment value at the current moment, its initial value is the output of the fuzzy logic inference engine, and the gain coefficient and During the offline calibration phase, a series of step response test signals are injected into the target hardware, and the value is determined with the goal of obtaining the critical damping response. This procedure transforms resource allocation from an open-loop request to a closed-loop control process with the actual response time as the goal.

[0041] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0042] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A display resource allocation system based on fuzzy logic, characterized in that: include: an interactive dynamic feature acquisition module configured to acquire a coordinate time series generated by user interaction at a sampling frequency of not less than 120 Hz, and calculate an acceleration vector sequence based on the coordinate time series; an intent authenticity identification unit configured to calculate a quantitative index representing the directional consistency of the acceleration vector sequence based on the acceleration vector sequence output by the interaction dynamic feature acquisition module within a time window, and to weight the acceleration amplitude of the interaction behavior according to the quantitative index; a fuzzy logic inference engine module configured to take the weighted interaction acceleration output by the intention authenticity discrimination unit as a first fuzzification input and, according to a fuzzy rule base stored in a memory, map the first fuzzification input into a resource allocation priority value during its operation cycle; A temperature sensing module, used to obtain a real-time temperature value of the system core processor; a cycle adaptive adjustment unit configured to dynamically adjust the operating cycle of the fuzzy logic inference engine module according to the real-time temperature value and a temperature and calculation delay characteristic model stored in the memory; A resource dynamic scheduling module is configured to allocate the system's computing resources according to the resource allocation priority value output by the fuzzy logic reasoning engine module.

2. A display resource allocation system based on fuzzy logic according to claim 1, characterized in that: The intention authenticity identification unit calculates a quantitative indicator of directional consistency. Specifically, it calculates the cosine value of the directional angle between each pair of consecutive acceleration vectors in the acceleration vector sequence within the time window; and performs arithmetic averaging on all calculated cosine values ​​to obtain an average cosine value as a quantitative indicator of directional consistency.

3. A display resource allocation system based on fuzzy logic according to claim 2, characterized in that: The intention authenticity identification unit generates the weighted interaction acceleration through the following relationship , ,in, is the original magnitude of the acceleration of the interaction behavior, is the quantitative index of directional consistency, namely the average cosine value. A data file stored in memory and A monotonically increasing weight function whose value range is from zero to one.

4. The display resource allocation system based on fuzzy logic according to claim 1, characterized in that: The cycle adaptive adjustment unit dynamically adjusts the operating cycle of the fuzzy logic inference engine module by determining an expected calculation delay based on the real-time temperature value and the temperature and calculation delay characteristic model; and subtracting the expected calculation delay from a baseline operating cycle to obtain a new and shortened operating cycle.

5. The display resource allocation system based on fuzzy logic according to claim 1, characterized in that: The fuzzy logic reasoning engine module is further configured to: take a value representing the content rendering complexity associated with the interactive behavior as a second fuzzification input, and combine the weighted interaction acceleration and the content rendering complexity values ​​to jointly determine the resource allocation priority value.

6. The display resource allocation system based on fuzzy logic according to claim 1, characterized in that: The fuzzy rule base includes the following rule: when the fuzzified value of the weighted interaction acceleration falls into a high-intensity interaction interval, the resource allocation priority value is determined to be a highest priority value. The execution of this rule is independent of the state of any other fuzzified input.

7. The display resource allocation system based on fuzzy logic according to claim 1, characterized in that: The system also includes a collaborative intention aggregation unit, which is configured to: confirm that a collaborative event has occurred when the interactive dynamic feature acquisition module detects at least two interactive behaviors with different spatial positions within a set time window, and the intention authenticity identification unit outputs high-intensity weighted interactive acceleration for at least two interactive behaviors; and based on the confirmation of the collaborative event, trigger a collaborative allocation strategy that concentrates computing resources in the area where at least two interactive behaviors are located.

8. The display resource allocation system based on fuzzy logic according to claim 1, characterized in that: The system also includes a cognitive load state tracking unit, which is configured to: calculate an exponential moving average representing the user's continuous cognitive load based on a series of resource allocation priority values ​​output by the fuzzy logic reasoning engine module over a period of time; and when the exponential moving average exceeds a load threshold stored in a memory, automatically adjust the presentation method of all notification information in the system except alarm information to reduce cognitive interference to the user.

Images