Mainboard active heat dissipation prediction control method based on load trend perception

Abstract

The present application relates to the technical field of computer hardware heat dissipation control, and particularly relates to a motherboard active heat dissipation prediction control method based on load trend perception. The method comprises the following steps: collecting the instantaneous electric parameters of the motherboard power supply loop and performing micro-time slot slicing processing, extracting the physical heat energy injected into the processor in each micro-time slot, and constructing an energy consumption pulse sequence reflecting the heat source energy injection trajectory; calculating the power rising trend of the energy consumption pulse sequence to obtain a trend characteristic quantity; continuously evaluating the trend characteristic quantity to obtain a load stability; and performing momentum synthesis operation on the load stability and the trend characteristic quantity to form a heat shock momentum value. By introducing the energy inventory management mechanism of the heat shock momentum and the virtual heat sink, the present application realizes the leap from passive temperature following to active energy defense, and significantly solves the contradiction between the heat accumulation risk under the transient high burst load and the silent experience.

Description

Technical Field

[0001] This invention relates to the field of computer hardware heat dissipation control technology, and in particular to a motherboard active heat dissipation predictive control method based on load trend perception. Background Technology

[0002] Current motherboard cooling technologies generally employ a simplified "static heat capacity model," treating the heatsink as a single entity with a constant heat absorption capacity. This ignores the significant time lag and resistance to heat transfer within the heatsink under transient high-frequency thermal shocks, constrained by the thermal conductivity and physical thermal resistance of the metal medium. When heat is still concentrated at the front of the conduction path, the heatsink at the far end has not yet participated in heat absorption. This leads to a severe overestimation of the system's current effective cooling capacity, an inability to accurately predict the risk of thermal breakdown, and consequently, an unexpected surge in core temperature.

[0003] Traditional technologies often rely on single-dimensional temperature feedback (response lag) or instantaneous current values ​​(high noise interference) for control. This approach cannot effectively distinguish between "short-term spurious glitches" and "real loads with continuous heating capabilities." This leads to the system frequently exhibiting erroneous aggressive cooling (generating noise) for transient fluctuations without actual thermal threat, or insufficient response to hidden, continuous heat accumulation (leading to heat buildup), making it difficult to achieve accurate momentum characterization.

[0004] Existing PID (Proportional-Integral-Derivative) temperature control strategies are mostly linear adjustment mechanisms, meaning the fan speed increases linearly with temperature. However, thermal saturation and thermal breakdown are often avalanche-like nonlinear processes. When the heat sink's thermal capacity is about to be breached, gentle linear adjustment cannot provide sufficient suppression, causing the optimal window for intervention to be missed just before the critical thermal saturation, resulting in the collapse of the heat dissipation defense system.

[0005] In summary, existing technologies generally suffer from problems such as distorted estimation of transient heat dissipation capacity, misjudgment of load response, and failure of thermal breakdown protection due to neglecting the dynamic characteristics of heat conduction, a single dimension of load identification, and linear control strategies. These problems urgently need to be solved. Summary of the Invention

[0006] Therefore, it is necessary to provide a motherboard active thermal prediction control method based on load trend perception to solve at least one of the above-mentioned technical problems.

[0007] To achieve the above objectives, a motherboard active thermal prediction control method based on load trend perception includes the following steps:

[0008] Step S1: Collect the instantaneous electrical parameters of the motherboard power supply circuit and perform micro-timeslot slicing processing to extract the physical thermal energy injected into the processor in each micro-timeslot and construct an energy consumption pulse sequence that reflects the energy injection trajectory of the heat source.

[0009] Step S2: Calculate the power rise trend of the energy consumption pulse sequence to obtain the trend characteristic quantity; perform a continuous evaluation of the trend characteristic quantity to obtain the load stability; perform momentum synthesis calculation on the load stability and the trend characteristic quantity to form the thermal shock momentum value.

[0010] Step S3: Obtain and initialize the radiator material parameters and evaluate the heat storage to obtain the radiator capacity value and the current heat storage; perform passive heat dissipation calculation based on the current heat storage to form the natural heat dissipation; optimize the heat flow path based on the natural heat dissipation and the current heat storage to generate a heat flow distribution map; use the radiator capacity value and the heat flow distribution map to assess the penetration risk of the thermal shock momentum value to obtain the heat penetration index.

[0011] Step S4: Use the preset dual-modal threshold to determine the threshold state of the heat penetration index and perform active gating intervention to obtain the heat dissipation gating command.

[0012] This invention eliminates the physical hysteresis of traditional temperature sensor feedback at the energy input source by constructing micro-timeslot slices and energy consumption pulse sequences for the power supply circuit. Utilizing thermal shock momentum analysis and stability assessment, it effectively filters noise caused by short-term transient disturbances, achieving accurate qualitative analysis of the load's true attack force. Furthermore, by combining a virtual heat sink model and heat flow path optimization, it accurately calculates the dynamic effective heat capacity and natural dissipation capacity of the heat sink while considering heat conduction hysteresis, maximizing the utilization of the physical heat absorption potential of the metal medium. Finally, through a dual-modal gating strategy, it forces silence in the heat capacity safety zone to utilize passive heat dissipation, and performs millisecond-level feedforward intervention in the heat capacity warning zone, thereby significantly reducing system operating noise and fan power consumption while ensuring the thermal safety of core hardware. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating a motherboard active thermal prediction control method based on load trend perception in this invention.

[0014] Figure 2 This is a schematic diagram of the hardware interaction architecture in this invention;

[0015] Figure 3 This is a schematic diagram of the core decision-making process of the heat dissipation strategy in this invention. Detailed Implementation

[0016] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0017] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0018] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0019] To achieve the above objectives, please refer to Figures 1 to 3 This invention provides a motherboard active thermal prediction control method based on load trend perception, comprising the following steps:

[0020] Step S1: Collect the instantaneous electrical parameters of the motherboard power supply circuit and perform micro-timeslot slicing processing to extract the physical thermal energy injected into the processor in each micro-timeslot and construct an energy consumption pulse sequence that reflects the energy injection trajectory of the heat source.

[0021] In this embodiment of the invention, the motherboard embedded controller synchronously latches the voltage and current register values ​​of the CPU power supply control chip via the system management bus to generate an aligned instantaneous power supply parameter set, and calculates the net effective power value through sliding window filtering. A high-precision hardware timer is read to obtain the actual sampling time width, and combined with the net effective power value and a preset packaged thermal shunt loss coefficient, the actual micro-slot thermal energy injection amount injected into the heat dissipation system within a single sampling period is calculated. The continuously generated injection amounts are stored in a first-in-first-out queue and encapsulated with timestamps and check bits to construct an energy consumption pulse sequence.

[0022] Step S2: Calculate the power rise trend of the energy consumption pulse sequence to obtain the trend characteristic quantity; perform a continuous evaluation of the trend characteristic quantity to obtain the load stability; perform momentum synthesis calculation on the load stability and the trend characteristic quantity to form the thermal shock momentum value.

[0023] In this embodiment of the invention, a first-order difference operation is performed on the energy consumption pulse sequence, and positively changing energy difference values ​​are selected to extract trend features including the rising slope and duration. A sliding observation window is constructed to statistically analyze the frequency of reverse fluctuations in the trend features, and a nonlinear normalization function is used to calculate the load stability, which characterizes the purity of the load waveform. The physical intensity slope in the trend features is weighted and multiplied with the load stability, and combined with the load change direction to generate a vectorized thermal shock momentum value that integrates the physical rate and the duration probability.

[0024] Step S3: Obtain and initialize the radiator material parameters and evaluate the heat storage to obtain the radiator capacity value and the current heat storage; perform passive heat dissipation calculation based on the current heat storage to form the natural heat dissipation; optimize the heat flow path based on the natural heat dissipation and the current heat storage to generate a heat flow distribution map; use the radiator capacity value and the heat flow distribution map to assess the penetration risk of the thermal shock momentum value to obtain the heat penetration index.

[0025] In this embodiment of the invention, a virtual heat storage model is constructed based on the physical parameters of the heat sink to track the current heat inventory. The natural heat dissipation, including radiation and convection components, is calculated by combining the ambient temperature and the Stefan-Boltzmann law. The effective conduction radius of the heat wave in the metallic medium is calculated based on the rise rate characteristics of the thermal shock momentum value, identifying the effective heat absorption area of ​​the heat sink and generating a heat flow distribution map. The total heat capacity of the heat sink is effectively converted using the heat flow distribution map to obtain a heat capacity margin value. The thermal shock momentum value is projected onto the fan response time window to predict the accumulated energy, and its ratio to the heat capacity margin value and the accumulated risk are calculated to generate a heat penetration index.

[0026] Step S4: Use the preset dual-modal threshold to determine the threshold state of the heat penetration index and perform active gating intervention to obtain the heat dissipation gating command;

[0027] In this embodiment of the invention, a dual-modal judgment logic is constructed using a preset thermal capacity safety threshold and a thermal capacity warning threshold. The thermal capacity penetration index is input to a hysteresis comparator to generate a working state value. When the state value is in the absorption state, the fan speed is forcibly locked to absorb heat using thermal capacity. When the state value is in the emission state, a high-speed target value is generated by looking up a table based on the index exceeding the limit. The priority of the target speed value and the default signals of the basic input / output system is analyzed, the control permission bit is set, and an overlay flag is encapsulated to generate a heat dissipation gating command to drive the fan to perform silent or strong cooling actions.

[0028] Preferably, step S1 includes:

[0029] The voltage register value and current register value of the CPU power supply control chip are synchronously locked through the motherboard system management bus to generate an instantaneous power supply parameter group.

[0030] Transient power synthesis and noise reduction are performed on the instantaneous power supply parameter set to obtain the net effective power value;

[0031] The net effective power value is integrated using micro-timeslot energy to obtain the micro-timeslot thermal energy injection amount;

[0032] A time-series queue is constructed and updated for the micro-timeslot thermal energy injection, forming an energy history cache queue;

[0033] The energy history cache queue is output as a discrete sequence normalization to generate an energy consumption pulse sequence.

[0034] In one embodiment, an embedded controller on the motherboard is used as the data acquisition core, establishing a communication connection with the CPU power supply control chip via a system management bus. At the start of each preset sampling period, the controller sends a synchronization latch instruction to freeze and read the values ​​of the voltage telemetry register and current telemetry register inside the power supply control chip within the same clock cycle. This synchronization operation eliminates phase offset errors caused by the time difference between voltage and current readings, generating a set of instantaneous power supply parameters containing precisely aligned voltage and current values.

[0035] Transient power synthesis is performed on the instantaneous power supply parameter group. The voltage and current values ​​within the group are multiplied to calculate the original instantaneous power value. To eliminate the electrical white noise of the sensor itself and nanosecond-level spikes without thermal effects, a sliding window weighted average algorithm is used to filter the original instantaneous power value. A sliding window of length 4 is established, with the current sampled value assigned a weight of 0.4, and the previous three historical sampled values ​​assigned weights of 0.3, 0.2, and 0.1 respectively. The weighted sum is then used to obtain the smoothed net effective power value.

[0036] Micro-timeslot energy integration is performed to convert electrical quantities into thermal quantities. The high-precision hardware timer value at the current sampling moment is read and subtracted from the timer value at the previous sampling moment to calculate the actual sampling duration between the two sampling times. The micro-timeslot thermal energy injection amount is set to... The formula for calculating the micro-timeslot thermal energy injection is:

[0037] ;

[0038] In the formula, Represents the net effective power value; Represents the actual sampling time width; This represents the preset packaged thermal shunt loss coefficient (e.g., 0.05). The micro-timeslot thermal energy injection amount is calculated using this formula, accurately quantifying the net physical thermal energy actually injected into the heat dissipation system and borne by it within a single sampling period.

[0039] Construct and maintain an energy history buffer queue. Allocate a fixed-length (e.g., N=40) first-in-first-out circular buffer in the controller's memory. Whenever a new micro-timeslot thermal energy injection is generated, push it into the head of the buffer while removing the earliest generated data from the tail. Attach a precise timestamp and data validity check bit to each data point in the queue, encapsulate it into a standard-format discrete data stream, and generate the final energy consumption pulse sequence.

[0040] Preferably, the micro-timeslot energy integration of the net effective power value includes:

[0041] The time slot boundaries are determined based on the net effective power value, resulting in the time slot interval value;

[0042] Power integration is performed on the time slot interval value and the net effective power value to obtain the original energy value;

[0043] The original energy value is corrected for heat loss to obtain the effective heat value;

[0044] The effective heat value is used to generate the injection amount, thus obtaining the micro-timeslot heat energy injection amount.

[0045] In one embodiment, time slot boundaries are determined and time slot interval values ​​are obtained. Using a high-frequency hardware timer built into the controller, the count value at the completion time of the current power sampling and the count value at the completion time of the previous power sampling are recorded respectively. The two count values ​​are subtracted and combined with the timer's clock frequency parameter to calculate the precise physical time span between the two sampling events, which is the time slot interval value. This operation eliminates sampling interval jitter caused by system interrupts or bus arbitration delays through real-time measurement.

[0046] A power integral operation is performed to obtain the raw energy value. Based on the Riemann sum approximation concept in calculus, the net effective power value is directly multiplied by the time slot interval value to calculate the total electrical energy consumed by the processor power supply circuit within that small time period, which is the raw energy value. This value represents the upper limit of the total electrical energy converted into heat.

[0047] Thermal loss correction is applied to the original energy value to eliminate ineffective heat components. Due to the physical packaging characteristics of the processor, some heat is conducted to the circuit board through the metal pins and does not participate in the heat exchange of the top heatsink. The package thermal resistance model pre-stored in the firmware is used to obtain the downward conduction ratio coefficient. The original energy value is multiplied by this coefficient to obtain the conduction loss value. The original energy value is subtracted from the conduction loss value to obtain only the heat energy conducted upward through the top cover, which is the effective heat value.

[0048] The final injection volume is generated based on the effective heat value. Considering the slight attenuation of heat flow due to the thermal resistance of the thermal interface material, a thermal interface transfer efficiency factor is introduced. Multiplying the effective heat value by this efficiency factor calculates the actual heat energy passing through the thermal interface material and absorbed by the radiator base, generating the micro-timeslot heat energy injection volume. This value precisely quantifies the net heat load that the cooling system must handle.

[0049] Preferably, the calculation of the power increase trend of the energy consumption pulse sequence in step S2 includes:

[0050] Pulse difference calculation is performed on the energy consumption pulse sequence to calculate the energy difference value characterizing the rate of change of energy injection;

[0051] The positively changing energy difference values ​​are selected to identify the rising phase of load power, and the rising slope and duration are extracted as trend features.

[0052] In one embodiment, pulse difference calculation is performed on the energy consumption pulse sequence. The data point P_t at the current sampling time and the data point P_{t-1} at the previous sampling time are extracted sequentially from the sequence. A first-order backward difference operation is performed, i.e., P_t is subtracted from P_{t-1}, and the result is divided by the sampling time interval Δt to calculate the energy difference value characterizing the rate of change of energy injection per unit time. This value reflects the rate and direction of change of load power at the current instant; a positive value represents an increase in power, and a negative value represents a decrease in power.

[0053] Filter out positive change values ​​and extract trend features. Iterate through the calculated energy difference values, setting a zero-value filter to retain only positive values ​​greater than zero, thus identifying the escalating phase of increasing load power. For consecutive positive energy difference values, their magnitude is directly recorded as the upward slope; simultaneously, a duration counter is started to count the number of consecutive positive changes, converting this into physical time length as the duration. The extracted upward slope value and the corresponding duration value are combined and encapsulated to generate a trend feature. This feature quantitatively describes the severity of the current load impact and the persistence of the acceleration process.

[0054] Preferably, the persistence assessment of trend characteristic quantities in step S2 includes:

[0055] The observation window is divided into observation intervals based on the trend characteristic quantities;

[0056] Extract fluctuation characteristic values ​​from the observed interval;

[0057] Persistence analysis is performed on the fluctuation characteristic values ​​to obtain persistence indicators;

[0058] The load stability is obtained by synthesizing the stability of the persistence index.

[0059] In one embodiment, an observation window is divided to determine the observation interval. A time-sliding first-in-first-out queue is constructed in memory as the observation window, with the window length set to cover the most recent N sampling periods. Whenever a new trend feature is generated, it is pushed onto the head of the queue and the data at the tail of the queue is discarded. All the continuous historical data currently stored in the queue constitutes the observation interval, establishing the time-domain range for evaluating load persistence.

[0060] Extract fluctuation characteristic values ​​from the observation interval. Traverse each data point within the observation interval and examine its energy change direction. Identify all data points with negative energy change direction (i.e., power decrease) or zero rate of change, and count the total number of times such data points appear within the interval. This count is used as the fluctuation characteristic value. This value quantifies the frequency of energy pullback or stagnation at the micro level within a macro-uptrend.

[0061] Persistence analysis is performed on the volatility characteristic values ​​to generate an indicator. Using a normalization algorithm, the volatility characteristic value is divided by the total number of samples N in the observation interval to calculate the proportion of reverse volatility. A purity conversion formula is defined, and this proportion is subtracted from the value 1 to obtain the persistence indicator. The closer this indicator value is to 1, the less reverse disturbance there is within the interval, and the more consistent the load increase process.

[0062] Stability synthesis is performed to output the final result. A nonlinear mapping function (such as a square function) is applied to the persistence index, with the persistence index as the independent variable input to the function. The calculated function value is used as the load stability. The formula for calculating load stability is:

[0063] ;

[0064] In the formula, For load stability, This represents the frequency of reverse fluctuations counted within the sliding observation window; This represents the total number of samples in the sliding observation window; This represents the nonlinear decay exponent (e.g., 2), used to adjust the sensitivity to signal jitter.

[0065] This synthesis process converts linear indices into probability coefficients that have high suppression characteristics against transient disturbances, accurately characterizing the stability of the current load waveform.

[0066] Preferably, step S2 includes performing momentum synthesis calculations on load stability and trend characteristics, which includes:

[0067] Extract the upward slope data from the trend characteristic quantity, and calculate the benchmark momentum by combining it with the load stability.

[0068] By performing a stability weight mapping on the reference momentum and load stability, the momentum weight value is obtained;

[0069] The load directionality is determined based on the momentum weight value, and a direction coefficient value is generated.

[0070] The thermal shock momentum value is generated based on the momentum weight value and the direction coefficient value.

[0071] In one embodiment, the rising slope is extracted and a reference momentum is calculated. The rising slope data, characterizing the instantaneous rate of change of power, is parsed from the trend features, and its absolute value is used as the basis for physical strength. This rising slope data is directly multiplied arithmetically with the load stability to obtain the reference momentum. This step uses the stability as a gain coefficient to perform preliminary amplitude modulation on the original slope.

[0072] A stability weight mapping is performed to obtain the momentum weight value. A piecewise mapping function is preset. When the load stability is below a preset noise threshold (e.g., 0.3), the output weight is forced to zero; when it is above the threshold, the output weight increases linearly with stability. The reference momentum and load stability are input into this mapping function to calculate the momentum weight value after nonlinear filtering. This operation achieves hard thresholding of low-probability transient glitches.

[0073] Perform load directionality determination to generate a direction factor value. Detect the sign of the energy difference value at the current moment. If the sign is positive, it indicates that the load is in the increasing phase, and the direction factor value is set to 1.0; if the sign is negative, it indicates that the load is in the decreasing phase, and the direction factor value is set to a rapid decay factor (e.g., 0.1) or zero. The direction factor value generated in this step is used to identify the effective direction of thermal shock and prevent negative momentum from interfering with prediction during the load decreasing phase.

[0074] The final thermal shock momentum value is generated. The momentum weight value is multiplied by the direction coefficient value to output the thermal shock momentum value. This value is a vector index that integrates the rate of physical change, waveform stability, and direction of change, accurately reflecting the attack momentum of the heat source posing a real thermal threat to the cooling system. The formula for calculating the thermal shock momentum value is:

[0075] ;

[0076] In the formula, This represents the thermal shock momentum value. This represents the absolute value of the power rise slope extracted from the trend characteristic quantity. The load stability calculated above; This represents the directional coefficient value (1 when the load is rising and 0 when it is falling).

[0077] Preferably, the passive heat dissipation calculation based on the current heat storage in step S3 includes:

[0078] Obtain ambient temperature data, and calculate the temperature difference coefficient based on the ambient temperature data and the current heat storage to obtain the temperature difference ratio value;

[0079] The radiative heat loss is estimated based on the temperature difference ratio to obtain the radiative heat loss value.

[0080] Calculate the convective heat dissipation value based on the temperature difference ratio and the radiative heat dissipation value.

[0081] The natural heat dissipation is generated by superimposing the radiative heat dissipation value and the convective heat dissipation value.

[0082] In one embodiment, a temperature difference coefficient is calculated to obtain a temperature difference ratio value. This is achieved by reading the values ​​from the motherboard edge cold zone sensor or by using a preset constant as the ambient temperature data. Based on the thermodynamic equation of state, the equivalent average surface temperature of the heatsink is calculated by dividing the current heat storage by the total physical heat capacity of the heatsink. Subtracting the ambient temperature data from this average surface temperature yields the instantaneous heat transfer temperature difference driving passive cooling. The ratio of this instantaneous heat transfer temperature difference to the ambient temperature is defined as the temperature difference ratio value, which characterizes the thermal potential energy gradient of the heatsink relative to the environment.

[0083] Perform a radiative heat dissipation estimation. According to the Stefan-Boltzmann law, the radiative heat power of a radiator is proportional to the fourth power of its surface temperature. Calculate the absolute temperature of the radiator surface using the temperature difference ratio, and combine this with preset surface emissivity constants and radiator surface area parameters to calculate the heat power released outwards in the form of electromagnetic waves, which is the radiative heat dissipation value.

[0084] Calculate the convective heat dissipation value. When the fan is off or running at low speed, airflow is primarily driven by the density gradient (buoyancy) caused by the temperature difference. Using the empirical formula for natural convection, the convective heat transfer coefficient is set to be proportional to the nth power of the temperature difference. The corrected convective heat transfer coefficient is multiplied by the effective convection area of ​​the radiator and the instantaneous heat transfer temperature difference to calculate the heat power carried away by natural airflow, which is the convective heat dissipation value.

[0085] Natural heat dissipation is generated. The radiative and convective heat dissipation values ​​are directly added together and multiplied by the duration of the current control cycle to calculate the total energy dissipated naturally by the heatsink during that time period. Considering that the enclosed environment inside the chassis can hinder heat dissipation, a boundary effect correction coefficient is introduced to adjust the total energy, ultimately outputting the natural heat dissipation. This value quantifies the self-healing capability of the cooling system. The formula for calculating natural heat dissipation is:

[0086] ;

[0087] In the formula, Natural heat dissipation Represents the Stefan-Boltzmann constant; Represents the emissivity of the radiator surface; Represents the effective radiation surface area; This represents the absolute surface temperature of the radiator, calculated from the current heat storage. Represents the ambient absolute reference temperature; Represents the natural convection heat transfer coefficient; Represents the effective convection area; This represents the duration of the current control cycle; This represents the chassis boundary effect correction factor (e.g., 0.85).

[0088] Preferably, the heat flow path optimization based on natural heat dissipation and current heat storage in step S3 includes:

[0089] Based on the natural heat dissipation and the current heat storage, a heat dissipation area analysis is performed to obtain a regional efficiency map.

[0090] A thermal resistance network is constructed using a regional efficiency map to obtain a thermal resistance distribution map;

[0091] Based on the thermal resistance distribution diagram, conduction path planning is performed to obtain the conduction path diagram;

[0092] A heat flow distribution map is generated based on the conduction path diagram.

[0093] In one embodiment, heat dissipation area analysis is performed to generate a regional efficiency map. The overall thermal saturation of the radiator is assessed using the ratio of natural heat dissipation to current heat storage. Based on the radiator's three-dimensional geometric model, the model is divided into several micro-element control volumes (such as the base area, heat pipe area, near-end fins, and far-end fins). The Euclidean distance from each micro-element to the center of the heat source is calculated according to the heat source location. Combined with the aforementioned thermal saturation level, each micro-element is assigned an initial efficiency weight: the closer the distance and the lower the saturation, the higher the weight. The generated mapping table containing spatial location and efficiency weights is the regional efficiency map.

[0094] A thermal resistance network is constructed to obtain a thermal resistance distribution map. Based on the regional efficiency map, the thermal conductivity parameter of the heat sink material is introduced. Each adjacent micro-element is considered as a node connected by thermal resistance, and the conduction thermal resistance between nodes is calculated. For paths containing heat pipes, extremely low thermal resistance values ​​are set; for pure aluminum fin paths, higher thermal resistance values ​​are set. A gridded data structure reflecting the resistance to heat diffusion from the base to the edge is constructed, which is the thermal resistance distribution map.

[0095] Plan the conduction path to generate a conduction path diagram. Combine the rise rate characteristics of the thermal shock momentum value generated in step S2 to calculate the wavefront velocity of the thermal wave under the current transient impact. Simulate the diffusion process of the heat wave from the heat source in the thermal resistance distribution diagram: the higher the impact velocity, the shorter the physical distance the heat wave can overcome thermal resistance to propagate per unit time. Mark all the node paths that the heat wave can actually reach within the current impact time window to form a conduction path diagram. This diagram defines the thermally active range under the current operating condition.

[0096] A heat flux distribution map is generated. Based on the conduction path map, the thermally active regions within the heat wave coverage area and the thermally blind regions outside the coverage area are identified. The proportion of the mass of each micro-element within the thermally active region to the total mass of the radiator is calculated, and combined with its weight in the regional efficiency map, a spatial weighted coefficient matrix is ​​generated. This matrix quantifies the effective heat absorption contribution of each physical part of the radiator under the current transient impact, which is the heat flux distribution map. This map is used in subsequent steps to effectively calculate the total physical heat capacity.

[0097] Preferably, step S3, which uses the radiator capacity value and heat flow distribution map to assess the penetration risk of thermal shock momentum value, includes:

[0098] The remaining heat capacity is estimated based on the radiator capacity, current heat storage, and heat flow distribution diagram to obtain the heat capacity margin value.

[0099] The thermal shock momentum value is normalized by the impact intensity to form a standard impact value;

[0100] The capacity ratio is obtained by calculating the capacity ratio between the standard impact value and the heat capacity margin value.

[0101] The cumulative effect of the capacity ratio is assessed to obtain the cumulative risk value;

[0102] The thermal penetration index is generated based on the cumulative risk value and the capacity ratio.

[0103] In one embodiment, residual heat capacity is estimated to obtain a heat capacity margin value. The radiator capacity (theoretical total physical heat capacity) is discounted using a spatial weighting coefficient matrix from the heat flow distribution map to calculate the effective total capacity that can actually participate in heat absorption within the current heat diffusion range. This effective total capacity is subtracted from the current heat storage, and the natural heat dissipation is added to calculate the physical energy difference between the radiator and its current instantaneous thermal saturation state; this is the heat capacity margin value. The formula for calculating the heat capacity margin value is:

[0104] ;

[0105] in, This is the heat capacity margin value. The total physical heat capacity of the radiator. This represents the mass percentage of the thermally active region based on the heat flow distribution map. This represents the current heat storage, which is the total amount of heat energy stored in the radiator at the current moment.

[0106] Impact intensity normalization is performed to form a standard impact value. A fan response time window parameter (e.g., 1.5 seconds) is set, corresponding to the physical inertia time required for the fan to accelerate from rest to its effective operating speed. The thermal impact momentum value (trend vector) is projected onto this time window, and the cumulative thermal energy to be injected into the cooling system in the future within this time period is predicted through time integration. The predicted cumulative energy is divided by a reference energy unit to convert it into a value consistent with the thermal capacity margin value, which is the standard impact value.

[0107] Calculate the capacity ratio. Divide the standard impact value by the thermal margin value. If the result is less than 1, it indicates that the predicted heat does not exceed the radiator's remaining heat absorption capacity; if the result is greater than 1, it indicates that the predicted heat will break through the radiator's physical capacity limit. This ratio directly reflects the power balance between the transient impact and the defensive margin, i.e., the capacity ratio.

[0108] A cumulative effect assessment is performed to obtain a cumulative risk value. The remaining time to thermal saturation is estimated using the capacity ratio. If the estimated time is less than the fan response time window, a risk weight is calculated using an exponential growth function to generate a higher cumulative risk value; if the estimated time is much greater than the response time window, a lower cumulative risk value is generated. This value quantifies the long-term overheating risk caused by continuous heat injection.

[0109] Generate the heat penetration index. The capacity ratio and cumulative risk value are weighted and synthesized. The capacity ratio is assigned a dominant weight (e.g., 0.7), and the cumulative risk value a secondary weight (e.g., 0.3). The weighted sum of the two is the heat penetration index. This index is a comprehensive judgment indicator used to indicate whether the current operating condition is about to breach the physical safety barrier of the heat dissipation system. The formula for calculating the heat penetration index is:

[0110] ;

[0111] In the formula, The heat penetration index is the thermal conductivity index. Represents the thermal shock momentum value; The term represents the preset fan response time window (e.g., 1.5 seconds), and the numerator represents the predicted projected energy. Represents the heat capacity margin value; This represents the cumulative risk value obtained based on the cumulative effect assessment.

[0112] Preferably, step S4 includes:

[0113] The heat penetration index is determined by a preset dual-modal threshold to obtain the working state value. The dual-modal threshold includes a heat capacity safety threshold and a heat capacity warning threshold.

[0114] A speed strategy is generated based on the working status value to form the target speed value;

[0115] Obtain the default control signal of the basic input / output system, and combine it with the target speed value to analyze the control priority and obtain the control authority bit.

[0116] The working status value, target speed value, and control authority bit are combined to form a heat dissipation gating command.

[0117] In one embodiment, a dual-modal threshold is used to determine the heat penetration index to generate an operating state value. A preset heat capacity safety threshold (e.g., 0.8) and a heat capacity warning threshold (e.g., 1.2) are established, and a hysteresis interval is constructed between them to prevent state jitter. The heat penetration index is input to a hysteresis comparator: if the index is lower than the heat capacity safety threshold, the system is determined to be in the heat capacity sufficient region, and a binary low level is output as the operating state value, indicating the absorption state; if the index is higher than the heat capacity warning threshold, the system is determined to be in the heat penetration danger region, and a binary high level is output as the operating state value, indicating the emission state; if the index is within the hysteresis interval, the state value of the previous moment remains unchanged.

[0118] The target speed is generated based on the operating status value. When the operating status value indicates absorption, the thermal capacity lock algorithm is activated, forcibly setting the target speed to the fan's minimum starting speed or zero speed, passively absorbing heat using the heat sink's thermal capacity. When the operating status value indicates discharge, the feedforward compensation algorithm is activated. Based on the extent to which the thermal penetration index exceeds the thermal capacity warning threshold, a preset nonlinear gain table is consulted to directly map and generate a high duty cycle PWM (Pulse-Width Modulation) value that is much higher than the speed corresponding to the current temperature, which is then used as the target speed value.

[0119] The control priority is analyzed to determine the control authority bit. The default temperature control PWM signal generated by the motherboard's basic input / output system is read in real time. The target speed value is compared with the amplitude of the default temperature control PWM signal. If the target speed value is greater than the default temperature control PWM signal and the operating state is in emission mode, or if the target speed value is less than the default temperature control PWM signal and the operating state is in absorption mode and the current core temperature is below the absolute safety threshold, then this control logic is determined to have higher priority, and the control authority bit is set (e.g., set to 1); otherwise, the control authority bit is reset (set to 0), allowing the basic input / output system to take over control.

[0120] Synthesize and output the gating instruction. Package the operating status value, target speed value, and control permission bit into a package. If the control permission bit is 1, generate an instruction package containing the override enable flag and the target speed value; if the control permission bit is 0, generate an instruction package containing the release flag. Send this instruction package to the fan control register of the Super I / O chip via the system management bus as the final thermal gating instruction, directly driving the physical fan to perform accelerated cooling or silent locking actions.

[0121] Please see Figure 2 This is a schematic diagram of the hardware interaction architecture, which shows the core linkage logic of "embedded controller-power supply chip-cooling fan": the embedded controller, as the core control unit, is connected to the power supply chip of the motherboard to collect its electrical parameters (for sensing load energy consumption), and is also connected to the cooling fan to output cooling control commands, realizing a closed-loop hardware link from sensing the load energy source to executing cooling, providing a hardware foundation for real-time response of active cooling.

[0122] Please see Figure 3 This is a schematic diagram of the core decision-making process of the heat dissipation strategy, corresponding to the thermal risk assessment and control link in the invention: The process takes "heat radiator capacity value" and "heat flow distribution thermal shock momentum value" as inputs, obtains quantified thermal risk indicators through "thermal penetration index assessment", and after combining with heat dissipation control commands, determines the heat dissipation state through "dual-mode gating decision" - when in the "absorption state", passive heat dissipation is relied on the heat radiator to ensure quiet operation, and when in the "emission state", active strong cooling is activated to avoid thermal breakdown, thus achieving a dynamic balance between heat dissipation efficiency and quiet experience.

[0123] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0124] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A motherboard active thermal prediction control method based on load trend perception, characterized in that, Includes the following steps: Step S1: Collect the instantaneous electrical parameters of the motherboard power supply circuit and perform micro-timeslot slicing processing to extract the physical thermal energy injected into the processor in each micro-timeslot and construct an energy consumption pulse sequence that reflects the energy injection trajectory of the heat source. Step S2: Calculate the power increase trend of the energy consumption pulse sequence and obtain the trend characteristic quantity; The load stability is obtained by continuously evaluating the trend characteristics; the load stability and trend characteristics are combined by momentum synthesis to form the thermal shock momentum value. Step S3: Obtain and initialize the radiator material parameters and evaluate the heat storage to obtain the radiator capacity value and current heat storage; calculate passive heat dissipation based on the current heat storage to form natural heat dissipation. Optimize the heat flow path based on natural heat dissipation and current heat storage to generate a heat flow distribution map; The thermal shock momentum value is used to assess the penetration risk using radiator capacity and heat flow distribution map, resulting in a thermal penetration index. This includes optimizing the heat flow path based on natural heat dissipation and current heat inventory. Based on the natural heat dissipation and the current heat storage, a heat dissipation area analysis is performed to obtain a regional efficiency map. A thermal resistance network is constructed using a regional efficiency map to obtain a thermal resistance distribution map; Based on the thermal resistance distribution diagram, conduction path planning is performed to obtain the conduction path diagram; Generate a heat flow distribution map based on the conduction path diagram; Step S4: Use the preset dual-modal threshold to determine the threshold state of the heat penetration index and perform active gating intervention to obtain the heat dissipation gating command.

2. The motherboard active heat dissipation predictive control method based on load trend perception according to claim 1, characterized in that, Step S1 includes: The voltage register value and current register value of the CPU power supply control chip are synchronously locked through the motherboard system management bus to generate an instantaneous power supply parameter group. Transient power synthesis and noise reduction are performed on the instantaneous power supply parameter set to obtain the net effective power value; The net effective power value is integrated using micro-timeslot energy to obtain the micro-timeslot thermal energy injection amount; A time-series queue is constructed and updated for the micro-timeslot thermal energy injection, forming an energy history cache queue; The energy history cache queue is output as a discrete sequence normalization to generate an energy consumption pulse sequence.

3. The motherboard active heat dissipation predictive control method based on load trend perception according to claim 2, characterized in that, Micro-timeslot energy integration of the net effective power value includes: The time slot boundaries are determined based on the net effective power value, resulting in the time slot interval value; Power integration is performed on the time slot interval value and the net effective power value to obtain the original energy value; The original energy value is corrected for heat loss to obtain the effective heat value; The effective heat value is used to generate the injection amount, thus obtaining the micro-timeslot heat energy injection amount.

4. The motherboard active heat dissipation predictive control method based on load trend perception according to claim 1, characterized in that, The power increase trend of the energy consumption pulse sequence calculated in step S2 includes: Pulse difference calculation is performed on the energy consumption pulse sequence to calculate the energy difference value characterizing the rate of change of energy injection; The positively changing energy difference values ​​are selected to identify the rising phase of load power, and the rising slope and duration are extracted as trend features.

5. The motherboard active heat dissipation predictive control method based on load trend perception according to claim 1, characterized in that, Step S2, which involves a persistent assessment of trend characteristics, includes: The observation window is divided into observation intervals based on the trend characteristic quantities; Extract fluctuation characteristic values ​​from the observed interval; Persistence analysis is performed on the fluctuation characteristic values ​​to obtain persistence indicators; The load stability is obtained by synthesizing the stability of the persistence index.

6. The motherboard active heat dissipation predictive control method based on load trend perception according to claim 1, characterized in that, Step S2 involves performing momentum synthesis calculations on load stability and trend characteristics, including: Extract the upward slope data from the trend characteristic quantity, and calculate the benchmark momentum by combining it with the load stability. By performing a stability weight mapping on the reference momentum and load stability, the momentum weight value is obtained; The load directionality is determined based on the momentum weight value, and a direction coefficient value is generated. The thermal shock momentum value is generated based on the momentum weight value and the direction coefficient value.

7. The motherboard active heat dissipation predictive control method based on load trend perception according to claim 1, characterized in that, Step S3, which involves passive heat dissipation calculations based on the current heat storage, includes: Obtain ambient temperature data, and calculate the temperature difference coefficient based on the ambient temperature data and the current heat storage to obtain the temperature difference ratio value; The radiative heat loss is estimated based on the temperature difference ratio to obtain the radiative heat loss value. Calculate the convective heat dissipation value based on the temperature difference ratio and the radiative heat dissipation value. The natural heat dissipation is generated by superimposing the radiative heat dissipation value and the convective heat dissipation value.

8. The motherboard active heat dissipation predictive control method based on load trend perception according to claim 1, characterized in that, Step S3, which uses the radiator capacity and heat flow distribution map to assess the penetration risk of thermal shock momentum, includes: The remaining heat capacity is estimated based on the radiator capacity, current heat storage, and heat flow distribution diagram to obtain the heat capacity margin value. The thermal shock momentum value is normalized by the impact intensity to form a standard impact value; The capacity ratio is obtained by calculating the capacity ratio between the standard impact value and the heat capacity margin value. The cumulative effect of the capacity ratio is assessed to obtain the cumulative risk value; The thermal penetration index is generated based on the cumulative risk value and the capacity ratio.

9. The motherboard active heat dissipation predictive control method based on load trend perception according to claim 1, characterized in that, Step S4 includes: The heat penetration index is determined by a preset dual-modal threshold to obtain the working state value. The dual-modal threshold includes a heat capacity safety threshold and a heat capacity warning threshold. A speed strategy is generated based on the working status value to form the target speed value; Obtain the default control signal of the basic input / output system, and combine it with the target speed value to analyze the control priority and obtain the control authority bit. The working status value, target speed value, and control authority bit are combined to form a heat dissipation gating command.

Images