An intelligent heat dissipation optimization management method and system for a multi-chip variable frequency driver

Abstract

This invention relates to the field of inverter heat dissipation management equipment technology, specifically to an intelligent heat dissipation optimization management method and system for multi-chip inverter drivers. The method involves collecting operating status parameters, thermal status parameters, and spatial distribution parameters of multiple power units to construct a heat diffusion priority propagation map for hotspot migration prediction; establishing a multi-fan anti-coupling interference matrix to describe the positive cooling effect and reverse disturbance effect of multiple cooling fans on different areas; constructing a local thermal risk constraint function; performing partitioned frequency misalignment joint optimization on the target speed, PWM duty cycle, and PWM switching frequency of multiple cooling fans to obtain the optimal fan control parameter combination; driving multiple cooling fans to operate; and performing online verification and parameter write-back correction based on the actual temperature distribution after control execution, hotspot migration results, and fan feedback results. This enables unified intelligent heat dissipation optimization management simultaneously addressing hotspot migration trends, wind field anti-coupling effects, and system energy consumption constraints.

Description

Technical Field

[0001] This invention relates to the field of inverter heat dissipation management equipment technology, specifically to an intelligent heat dissipation optimization management method and system for a multi-chip inverter driver. Background Technology

[0002] Variable frequency drive (VFD) cooling systems primarily rely on temperature threshold-triggered control, experience-based speed regulation, or single-loop closed-loop speed control, typically increasing the speed of a single fan directly based on the temperature at a local measurement point. While this approach is applicable in single-heat-source or weakly coupled scenarios, it becomes problematic in multi-chip VFDs where thermal coupling is prevalent among multiple power units. Furthermore, overlapping airflow paths, airflow competition, and localized backflow issues among multiple fans lead to significant multivariable, strongly coupled, and nonlinear characteristics. Traditional methods struggle to accurately identify hotspot migration trends and determine whether increasing the speed of a single fan will trigger reverse heating in adjacent areas, enhance local disturbances, or increase overall energy consumption.

[0003] From the perspective of market and engineering application needs, users not only require drivers to maintain safe junction temperatures under high load and long-term operating conditions, but also require reduced fan energy consumption, reduced noise, extended device lifespan, and improved overall system reliability. Therefore, there is an urgent need for an intelligent thermal management method that can predict, evaluate, and perform closed-loop optimization for multi-chip thermal diffusion, multi-fan cooperative interference, and dynamic operating condition changes, in order to meet the technical requirements of high-performance inverter drivers for refined, real-time, and adaptive thermal control. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an intelligent heat dissipation optimization management method and system for multi-chip inverter drivers. It aims to address the problems of significant thermal coupling among multiple power units within a multi-chip inverter driver, mutual interference between airflows from multiple fans, and the difficulty in achieving global collaborative optimization by traditional control methods that can only passively adjust based on local temperatures. The invention can simultaneously address hotspot migration trends, airflow anti-coupling effects, and system energy consumption constraints through unified intelligent heat dissipation optimization management.

[0005] This invention provides an intelligent heat dissipation optimization management method for a multi-chip frequency converter driver, including step S10, collecting the operating status parameters, thermal status parameters and spatial distribution parameters of multiple power units, and constructing a heat diffusion priority propagation map for hot spot migration prediction by combining the thermal conduction correlation between power units;

[0006] Step S20: Based on the heat diffusion priority propagation map constructed in step S10, and combined with the installation positions of multiple cooling fans, air duct coverage relationship, and air supply overlap relationship, a multi-fan anti-coupling interference matrix is ​​established to describe the positive cooling effect and reverse disturbance effect of multiple cooling fans on different areas.

[0007] Step S30: Based on the heat diffusion priority propagation diagram described in step S10 and the multi-fan anti-coupling interference matrix described in step S20, a local thermal risk constraint function is constructed to screen out pseudo-optimal control schemes that can reduce the local target temperature but will cause the adjacent area temperature to deteriorate.

[0008] Step S40: Under the constraint of the local thermal risk function described in step S30, the target speed, PWM duty cycle and PWM switching frequency of multiple cooling fans are jointly optimized by partitioning and frequency shifting to obtain the optimal fan control parameter combination that satisfies the junction temperature limit of each power unit, reduces the highest hot spot temperature and suppresses the synchronous beat effect of multiple fans.

[0009] In step S50, multiple cooling fans are driven to run according to the optimal fan control parameter combination obtained in step S40. Based on the actual temperature distribution, hot spot migration results and fan feedback results after control execution, online verification and parameter write-back correction are performed on the heat diffusion priority propagation map in step S10 and the multi-fan anti-coupling interference matrix in step S20.

[0010] In one embodiment, step S10 includes:

[0011] Step S11: Collect the operating current, operating voltage, switching status, loss estimate, case temperature or near-junction temperature, local ambient temperature and load rate information of multiple power units, and obtain the spatial coordinates, adjacency relationship, common heat dissipation structure relationship and heat conduction path relationship of each power unit inside the driver.

[0012] Step S12: Calculate the priority weight of the thermal influence propagated from the i-th power unit to the j-th power unit based on the equivalent heating power, temperature rise rate, thermal structure correlation coefficient between the i-th power unit and the j-th power unit, and thermal path length.

[0013] Step S13: Establish a directed heat propagation relationship from the source power unit to the target power unit according to the priority weight between each power unit, and construct a heat diffusion priority propagation map for hot spot migration prediction based on each directed heat propagation relationship.

[0014] In one embodiment, step S20 includes:

[0015] Step S21: Based on the heat diffusion priority propagation map obtained in step S10, the regions where multiple power units are located are dynamically divided to obtain the hot spot core area, hot spot diffusion front area, ordinary heating area and reverse cooling sensitive area.

[0016] Step S22: Obtain the installation position, air outlet direction, rated air volume, current speed, PWM drive parameters, air duct coverage boundary, and relative positional relationship between multiple cooling fans;

[0017] Step S23: Calculate the overall interference intensity of the m-th cooling fan on the n-th region based on the lateral turbulence intensity, backflow effect intensity, and actual effective cooling contribution generated by the m-th cooling fan on the n-th region.

[0018] Step S24: Based on the comprehensive interference intensity of each cooling fan in each region, establish the mapping relationship of the effects of multiple cooling fans on different regions, and extract the effective cooling contribution of each cooling fan to the target region, the lateral turbulence influence on adjacent regions, the backflow effect on adjacent air ducts, and the weakening effect on local air supply uniformity, to form a multi-fan anti-coupling interference matrix.

[0019] In one embodiment, step S30 includes:

[0020] Step S31: Based on the heat diffusion priority propagation map output in step S10, extract the hot spot core region, the hot spot migration front region, and the candidate affected region that has a heat propagation relationship with the hot spot migration front region from multiple power units.

[0021] Step S32: Based on the multi-fan anti-coupling interference matrix output in step S20, extract the positive cooling component and negative disturbance component of each cooling fan to the hot spot core region, the hot spot migration front region and the candidate affected region.

[0022] Step S33: Construct a local thermal risk constraint function based on the predicted maximum junction temperature of all power units, the positive temperature rise increment of non-target areas, the comprehensive interference intensity of each cooling fan on different areas, and the correlation of the fan's effect on the corresponding area.

[0023] Step S34: Based on the local thermal risk constraint function, perform risk screening on the candidate fan control schemes and eliminate pseudo-optimal control schemes that, although they can reduce the local target temperature, may cause reverse heating in adjacent areas, accelerated hot spot migration, or enhanced local disturbances.

[0024] In one embodiment, step S40 includes:

[0025] Step S41: Based on the evaluation of the local thermal risk constraint function constructed in step S30, the target speed, PWM duty cycle and PWM switching frequency of multiple cooling fans are defined as the control variables to be optimized, and the frequency interval relationship, phase stagger relationship and regional primary and secondary coverage relationship between multiple fans are incorporated into a unified parameter space.

[0026] Step S42: Based on the hot spot diffusion direction indicated by the heat diffusion priority propagation map described in step S10, prioritize the allocation of fan adjustment authority and adjustment range for the corresponding area, and restrict the synchronous speed-up behavior and same-frequency drive behavior of adjacent fans according to the multi-fan anti-coupling interference matrix described in step S20.

[0027] Step S43: Jointly evaluate the local thermal risk target value corresponding to the candidate control parameter combination, the total power consumption of all cooling fans, and the synchronous beat intensity index of multiple cooling fans to obtain the comprehensive optimization target value;

[0028] Step S44: Under the conditions of meeting the upper limit of junction temperature of each power unit, the allowable air loss boundary of the system and the local disturbance tolerance, perform partitioned frequency misalignment joint optimization for multiple cooling fans to obtain the optimal fan control parameter combination that minimizes the comprehensive optimization objective value.

[0029] In one embodiment, step S50 includes:

[0030] Step S51: Based on the optimal fan control parameter combination output in step S40, send corresponding target speed commands, PWM duty cycle commands and PWM switching frequency commands to multiple cooling fans respectively, so that multiple cooling fans can operate in coordination according to a predetermined partitioned frequency misalignment control method.

[0031] Step S52: After control execution, continuously collect the actual temperature value, temperature rise rate value, hot spot location change result, hot spot migration path change result, actual speed value of each cooling fan, feedback current value, local wind speed feedback value, and air duct pressure difference feedback value of each power unit.

[0032] Step S53: Calculate the model correction trigger amount based on the actual temperature, predicted temperature, actual hot spot migration status, and predicted hot spot migration status of the power unit.

[0033] Step S54: When the model correction trigger amount is greater than the preset correction threshold, the edge weight parameters, propagation priority parameters and region division parameters of the heat diffusion priority propagation map in step S10 are written back and corrected, and the cooling contribution parameters, lateral turbulence parameters and backflow effect parameters of the multi-fan anti-coupling interference matrix in step S20 are corrected synchronously.

[0034] In step S55, the corrected heat diffusion priority propagation map and multi-fan anti-coupling interference matrix are re-output to steps S30 and S40 to form a closed-loop dynamic heat dissipation optimization management process.

[0035] This invention provides an intelligent heat dissipation optimization management system for a multi-chip inverter driver, comprising:

[0036] The data acquisition module is used to collect the operating status parameters, thermal status parameters, and spatial distribution parameters of multiple power units, as well as the installation position, air duct coverage relationship, air supply overlap relationship, and feedback operation information of multiple cooling fans.

[0037] The heat diffusion construction module is used to receive the operating status parameters, thermal status parameters and spatial distribution parameters of multiple power units transmitted by the data acquisition module, and construct a heat diffusion priority propagation map for hot spot migration prediction by combining the thermal conduction correlation between power units, and dynamically update the heat diffusion priority propagation map.

[0038] The anti-coupling interference matrix module receives the heat diffusion priority propagation map transmitted by the heat diffusion construction module, and combines the installation positions of multiple cooling fans, air duct coverage relationship, and air supply overlap relationship to establish a multi-fan anti-coupling interference matrix to describe the positive cooling effect and reverse disturbance effect of multiple cooling fans on different areas, and is used to dynamically update the anti-coupling interference matrix.

[0039] The local thermal risk constraint module is used to receive the thermal diffusion propagation map transmitted by the thermal diffusion construction module, receive the anti-coupling interference matrix transmitted by the anti-coupling interference matrix module, construct the local thermal risk constraint function, and screen out pseudo-optimal control schemes that can reduce the local target temperature but will cause the adjacent area temperature to deteriorate.

[0040] The partitioned frequency optimization module is used to receive the local thermal risk constraint function transmitted by the local thermal risk constraint module, and perform partitioned frequency joint optimization on the target speed, PWM duty cycle and PWM switching frequency of multiple cooling fans based on the local thermal risk constraint function, so as to obtain the optimal fan control parameter combination that satisfies the junction temperature limit of each power unit, reduces the highest hot spot temperature and suppresses the synchronous beat effect of multiple fans.

[0041] The execution module is used to drive multiple cooling fans to operate according to the optimal combination of fan control parameters;

[0042] The real-time monitoring module is used to collect actual temperature distribution data, hot spot migration data, and fan feedback data after the execution module is executed.

[0043] The online verification and correction module is used to acquire the actual temperature distribution data, hot spot migration data and fan feedback data transmitted by the real-time monitoring module, and feed the data back to the heat diffusion construction module and the anti-coupling interference matrix module for online verification and parameter write-back correction. Attached Figure Description

[0044] Figure 1 One of the flowcharts for an intelligent heat dissipation optimization management method for a multi-chip frequency converter provided by the present invention;

[0045] Figure 2 The second flowchart illustrates an intelligent heat dissipation optimization management method for a multi-chip frequency converter provided by the present invention.

[0046] Figure 3The third flowchart illustrates an intelligent heat dissipation optimization management method for a multi-chip frequency converter provided by the present invention.

[0047] Figure 4 The fourth flowchart illustrates an intelligent heat dissipation optimization management method for a multi-chip frequency converter provided by the present invention.

[0048] Figure 5 The fifth flowchart illustrates an intelligent heat dissipation optimization management method for a multi-chip frequency converter provided by the present invention.

[0049] Figure 6 The sixth flowchart illustrates the intelligent heat dissipation optimization management method for a multi-chip frequency converter provided by this invention.

[0050] Figure 7 This is a schematic diagram of the structure of an intelligent heat dissipation optimization management system for a multi-chip frequency converter provided by the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] The following is combined with Figures 1 to 7 This invention describes an intelligent heat dissipation optimization management method and system for a multi-chip frequency converter driver.

[0053] like Figure 1 As shown, in one embodiment, an intelligent heat dissipation optimization management method for a multi-chip inverter driver includes the following steps:

[0054] Step S10: Collect the operating status parameters, thermal status parameters, and spatial distribution parameters of multiple power units, and construct a heat diffusion priority propagation map for hot spot migration prediction by combining the thermal conduction correlation between power units;

[0055] In multi-chip inverter drivers, besides the current highest temperature point itself during thermal runaway, new hotspot chains can also form from this hotspot to the surrounding areas after a sudden load change. If the control strategy only focuses on the current highest temperature point, it is easy to encounter the problem that the hotspot has already migrated and the fan operation fails when the control response arrives. Therefore, this step establishes a heat diffusion priority propagation map by integrating power loss, temperature rise rate, and geometric adjacency relationship, which is used to characterize the strength of the heat propagation tendency of each power unit to its neighboring units.

[0056] Step S20: Based on the heat diffusion priority propagation map constructed in step S10, and combined with the installation positions of multiple cooling fans, air duct coverage relationship, and air supply overlap relationship, establish a multi-fan anti-coupling interference matrix to describe the positive cooling effect and reverse disturbance effect of multiple cooling fans on different areas.

[0057] When a fan accelerates, it can cause neighboring areas to become hotter due to airflow competition, lateral recirculation, localized negative pressure, or airflow obstruction. This problem is particularly prominent in multi-fan parallel or counter-flow configurations and is difficult to address using conventional single-fan control methods. Therefore, this step establishes an anti-coupling interference matrix to specifically characterize the disturbance effect of a fan's enhanced operation on its non-target areas, thereby identifying which areas belong to the "reverse cooling sensitive zone".

[0058] Step S30: Based on the heat diffusion priority propagation diagram described in step S10 and the multi-fan anti-coupling interference matrix described in step S20, a local thermal risk constraint function is constructed, which includes the maximum junction temperature constraint, the neighboring region reverse temperature rise constraint, the local interference penalty constraint and the system energy consumption constraint. This function is used to screen out pseudo-optimal control schemes that can reduce the local target temperature but will cause the temperature rise of the adjacent region to deteriorate.

[0059] Step S40: Under the constraint of the local thermal risk function described in step S30, the target speed, PWM duty cycle and PWM switching frequency of multiple cooling fans are jointly optimized by partitioning and frequency shifting to obtain the optimal fan control parameter combination that satisfies the junction temperature limit of each power unit, reduces the highest hot spot temperature and suppresses the synchronous beat effect of multiple fans.

[0060] In this step, under the joint constraints of the heat propagation diagram of S1 and the local risk function of S3, the speed and duty cycle of multiple fans are jointly optimized. The PWM frequency misalignment is also incorporated into the control variables so that multiple fans no longer run at the same frequency and rhythm, thereby reducing airflow oscillation, local pulsation superposition and flow field periodic instability.

[0061] In step S50, multiple cooling fans are driven to run according to the optimal fan control parameter combination obtained in step S40. Based on the actual temperature distribution, hot spot migration results and fan feedback results after control execution, online verification and parameter write-back correction are performed on the heat diffusion priority propagation map in step S10 and the multi-fan anti-coupling interference matrix in step S20 to form a closed-loop dynamic heat dissipation optimization management process for multi-chip thermal coupling scenarios.

[0062] If the actual heat migration direction is inconsistent with the prediction of S1, or if the wind field interference deviates from the estimate of S2, then it is necessary to reverse-correct the heat propagation map and interference matrix.

[0063] To address the issues of significant thermal coupling among multiple power units in existing multi-chip inverter drivers, mutual interference between airflow from multiple fans, and the difficulty of balancing hotspot suppression and system power consumption using traditional temperature control methods, this invention first collects the operating status, temperature status, and spatial distribution information of multiple power units to construct a heat diffusion priority propagation map for hotspot migration prediction. Further, combining the installation positions of multiple fans, airflow coverage relationships, and operational feedback, a multi-fan anti-coupling interference matrix is ​​established to identify the cooling contribution and disturbance side effects of fans to different areas. Based on this, a local thermal risk constraint function is constructed, including maximum junction temperature constraints, neighboring area reverse temperature rise constraints, and airflow interference penalties, to screen candidate control schemes. Subsequently, a partitioned frequency misalignment joint optimization is performed on the target speeds of multiple fans, PWM duty cycles, and PWM switching frequencies to obtain the optimal control parameter combination that satisfies junction temperature limits while balancing energy consumption and flow field stability. Finally, based on actual temperature feedback and hotspot migration results, the heat diffusion priority propagation map and the multi-fan anti-coupling interference matrix are verified and their parameters corrected online, forming a closed-loop adaptive heat dissipation optimization mechanism. This invention can effectively reduce the maximum junction temperature of multi-chip inverter drivers, suppress the reverse temperature rise phenomenon in adjacent areas, reduce beat vibration and redundant air loss in multi-fan coordinated operation, and improve the accuracy, stability and real-time adaptability of heat dissipation control under complex thermal coupling conditions.

[0064] like Figure 2 As shown, in one embodiment, the intelligent heat dissipation optimization management method for a multi-chip frequency converter provided by the present invention includes the following steps in step S10:

[0065] Step S11: Collect the operating current, operating voltage, switching status, loss estimate, case temperature or near-junction temperature, local ambient temperature and load rate information of multiple power units, and obtain the spatial coordinates, adjacency relationship, common heat dissipation structure relationship and heat conduction path relationship of each power unit inside the driver.

[0066] Step S12: Based on the equivalent heating power Q(i,t), temperature rise rate G(i,t), thermal structure correlation coefficient A(i,j) between the i-th power unit and the j-th power unit, and thermal path length L(i,j), calculate the priority weight W(i,j,t) for the thermal influence propagated from the i-th power unit to the j-th power unit. The calculation formula is as follows:

[0067] W(i,j,t)=k1*Q(i,t)*A(i,j)+k2*G(i,t)+k3 / L(i,j);

[0068] Where W(i,j,t) represents the priority weight of the thermal influence propagating from the i-th power unit to the j-th power unit at time t; Q(i,t) represents the equivalent heating power of the i-th power unit at time t; A(i,j) represents the thermal structure correlation coefficient between the i-th power unit and the j-th power unit; G(i,t) represents the rate of temperature change of the i-th power unit at time t; L(i,j) represents the thermal path length between the i-th power unit and the j-th power unit; and k1, k2, and k3 represent weighting coefficients.

[0069] This formula characterizes the priority of heat diffusion from the i-th power unit to the j-th power unit, unifying heat intensity, temperature rise trend, and ease of propagation into a single weight value: the greater the heat generated by the i-th power unit, the faster its temperature rise, and the shorter the thermal path between it and the j-th power unit, the larger W(i,j,t) will be, indicating that this direction is more likely to become the main channel for hotspot migration. The purpose of this formula is not to calculate the final temperature, but rather to identify the possible directions of hotspot diffusion in advance, providing a predictive basis for subsequent fan zoning control and the determination of key cooling areas.

[0070] Step S13: Establish a directed heat propagation relationship from the source power unit to the target power unit according to the priority weight W(i,j,t) between each power unit, and construct a heat diffusion priority propagation map for hot spot migration prediction based on each directed heat propagation relationship, so as to characterize the directionality, priority and propagation intensity of the hot spot expanding from the current high heat area to the potentially affected area.

[0071] The heat diffusion priority propagation map is output to step S20 as the input basis for subsequent fan action area division, hot spot diffusion front identification and key cooling area determination.

[0072] like Figure 3 As shown in one embodiment, the intelligent heat dissipation optimization management method for a multi-chip inverter driver provided by the present invention includes the following steps in step S20:

[0073] Step S21: Based on the heat diffusion priority propagation map obtained in step S10, the regions where multiple power units are located are dynamically divided to obtain the hot spot core area, the hot spot diffusion front area, the ordinary heating area, and the reverse cooling sensitive area.

[0074] Step S22: Obtain the installation position, air outlet direction, rated air volume, current speed, PWM drive parameters, air duct coverage boundary, and relative positional relationship between multiple cooling fans.

[0075] Step S23, based on the lateral turbulence intensity V generated by the m-th cooling fan on the n-th region. cross (m,n,t), intensity of backflow or backpressure effect P back(m,n,t) and the actual effective cooling contribution C cool Given (m,n,t), calculate the overall interference intensity D(m,n,t) of the m-th cooling fan on the n-th region. The calculation formula is:

[0076] D(m,n,t)=h1*V cross (m,n,t)+h2*P back (m,n,t)-h3*C cool (m,n,t);

[0077] Where D(m,n,t) represents the overall interference intensity of the m-th cooling fan on the n-th region at time t; V cross (m,n,t) represents the lateral turbulence intensity caused by the m-th cooling fan to the n-th region; P back (m,n,t) represents the intensity of the backflow or backpressure effect caused by the m-th cooling fan on the n-th region; C cool (m,n,t) represents the actual effective cooling contribution of the m-th cooling fan to the n-th region; h1, h2, and h3 represent the proportionality coefficients of the corresponding terms.

[0078] This formula is used to evaluate the overall interference intensity of the m-th fan on the n-th region. Its modeling idea is to simultaneously consider the negative disturbances and positive cooling caused by fan operation: lateral turbulence and backflow pressure exacerbate local wind field imbalance, thus being treated as positive additive terms; actual effective cooling weakens the overall interference, thus being treated as a subtractive term. By determining whether the fan operation is net beneficial or net harmful to the target region and neighboring regions, a quantitative basis is provided for identifying reverse cooling sensitive areas and constructing an anti-coupling interference matrix.

[0079] Step S24: Based on the comprehensive interference intensity D(m,n,t) of each cooling fan in each region, establish the mapping relationship of the effects of multiple cooling fans on different regions, and extract the effective cooling contribution of each cooling fan to the target region, the lateral turbulence influence on adjacent regions, the backflow effect on adjacent air ducts, and the weakening effect on local air supply uniformity, to form a multi-fan anti-coupling interference matrix.

[0080] The multi-fan anti-coupling interference matrix is ​​output to step S30 as the perturbation input for constructing local thermal risk constraints.

[0081] like Figure 4 As shown, in one embodiment, the intelligent heat dissipation optimization management method for a multi-chip frequency converter provided by the present invention includes the following steps in step S30:

[0082] Step S31: Based on the heat diffusion priority propagation map output in step S10, extract the hot spot core region, the hot spot migration front region, and the candidate affected region that has a heat propagation relationship with the hot spot migration front region from multiple power units.

[0083] Step S32: Based on the multi-fan anti-coupling interference matrix output in step S20, extract the positive cooling component and negative disturbance component of each cooling fan to the hot spot core region, the hot spot migration front region and the candidate affected region.

[0084] Step S33, based on the predicted maximum junction temperature max[T] of all power units j [i,t)], positive temperature rise increment ΔT in the non-target region plus The local thermal risk constraint function J is constructed by considering the comprehensive interference intensity D(m,z,t) of each cooling fan on different areas and the correlation quantity B(m,z,t) of the fan's effect on the corresponding area. risk (t), the calculation formula is:

[0085] ,

[0086] in, T represents the target value of local thermal risk corresponding to a candidate control scheme at time t; j (i,t) represents the predicted junction temperature of the i-th power unit at time t; max[T j [i,t)] represents the highest predicted junction temperature among all power cells; ΔT plus (z,t) represents the positive temperature rise increment of the z-th non-target region; D(m,z,t) represents the comprehensive interference intensity of the m-th cooling fan on the z-th region; B(m,z,t) represents the correlation quantity of the effect of the m-th cooling fan on the z-th region; p1 and p2 represent the penalty coefficients.

[0087] This formula is used to construct the local thermal risk value of candidate control schemes. It simultaneously incorporates the risk of the hottest spot, the risk of warming in neighboring areas, and the side effects of fan disturbances into the evaluation. The first term ensures that the current highest temperature point is constrained; the second term prevents a control action from lowering the local temperature but causing warming in other areas; and the third term reflects the collateral disturbance effects of fan actions on different areas. This formula allows pseudo-optimal schemes that could cause reverse warming in neighboring areas or deterioration of the local wind field to be screened out in advance, ensuring that subsequent optimization is based on more reliable risk constraints.

[0088] Step S34, based on the local thermal risk constraint function Risk screening was conducted on candidate fan control schemes to eliminate pseudo-optimal control schemes that, while able to reduce the local target temperature, would cause reverse heating in adjacent areas, accelerated hot spot migration, or enhanced local disturbances.

[0089] The candidate control schemes that pass the screening are output to step S40 as a restricted search space for multi-fan joint optimization.

[0090] like Figure 5 As shown, in one embodiment, the intelligent heat dissipation optimization management method for a multi-chip frequency converter provided by the present invention includes the following steps in step S40:

[0091] Step S41: Based on the evaluation of the local thermal risk constraint function constructed in step S30, the target speed, PWM duty cycle and PWM switching frequency of multiple cooling fans are defined as the control variables to be optimized, and the frequency interval relationship, phase stagger relationship and regional primary and secondary coverage relationship between multiple fans are incorporated into a unified parameter space.

[0092] Step S42: Based on the hot spot diffusion direction indicated by the heat diffusion priority propagation map described in step S10, prioritize the allocation of fan adjustment authority and adjustment range for the corresponding area, and restrict the synchronous speed-up behavior and same-frequency drive behavior of adjacent fans according to the multi-fan anti-coupling interference matrix described in step S20.

[0093] Step S43, for the local thermal risk target value J corresponding to the candidate control parameter combination. risk (t), Total power consumption of all cooling fans P fan (t) and the synchronous beat intensity index S of multiple cooling fans sync (t) Conduct a joint evaluation to obtain the comprehensive optimization target value J. opt The calculation formula is:

[0094] J opt =q1*J risk (t)+q2*P fan (t)+q3*S sync (t);

[0095] Among them, J opt J represents the comprehensive optimization objective value corresponding to the candidate control parameter combination; risk (t) represents the target value of local thermal risk at time t; P fan (t) represents the total power consumption of multiple cooling fans at time t; S sync (t) represents the synchronous beat intensity index of multiple cooling fans under the current control scheme; q1, q2, and q3 represent the weight coefficients of each evaluation item;

[0096] J risk (t) Ensure thermal safety, P fan (t) Constrains fan energy consumption, S sync(t) is used to suppress flow field resonance and pulsation superposition caused by multiple fans operating at the same or near-the same frequency. Its optimization results can not only reduce hot spot temperature, but also reduce redundant air consumption and improve the quality of multi-fan cooperative operation.

[0097] Step S44: Using an intelligent iterative optimization method, under the conditions of meeting the upper limit of junction temperature of each power unit design, the system allowable air loss boundary, and the local disturbance tolerance, a partitioned frequency misalignment joint optimization is performed on multiple cooling fans to obtain the comprehensive optimization target value J. opt Minimum optimal fan control parameter combination X best Specifically, this intelligent iterative optimization method is used to optimize the collaborative operation of multiple fans. It can utilize swarm intelligence algorithms such as particle swarm optimization, or adopt genetic algorithms, differential evolution algorithms, model predictive control, or reinforcement learning methods to jointly solve for fan speed, PWM duty cycle, and frequency parameters. Among these, if the control computing power and sample conditions are sufficient, the scheme based on model prediction or reinforcement learning may achieve better real-time control performance under dynamic operating conditions.

[0098] The optimal fan control parameter combination X best The output is sent to step S50 as the execution input for the coordinated driving of multiple cooling fans.

[0099] like Figure 6 As shown, in one embodiment, the intelligent heat dissipation optimization management method for a multi-chip inverter driver provided by the present invention includes the following steps in step S50:

[0100] Step S51, based on the optimal fan control parameter combination X output in step S40 best The system sends corresponding target speed commands, PWM duty cycle commands, and PWM switching frequency commands to multiple cooling fans, enabling the multiple cooling fans to operate collaboratively according to a predetermined partitioned frequency misalignment control method.

[0101] Step S52: After control execution, continuously collect the actual temperature value, temperature rise rate value, hot spot location change result, hot spot migration path change result, actual speed value of each cooling fan, feedback current value, local wind speed feedback value, and air duct pressure difference feedback value of each power unit.

[0102] Step S53, based on the actual temperature T of the i-th power unit real (i,t), predicted temperature T pre (i,t), actual hotspot migration status M real (t) and predicted hotspot migration status M pre (t), calculate the model correction trigger value E update (t), the calculation formula is:

[0103] E update(t)=u1*|T real (i,t)-T pre (i,t)|+u2*|M real (t)-M pre (t)|;

[0104] Among them, E update (t) represents the model correction trigger value at time t; T real (i,t) represents the actual measured temperature of the i-th power unit at time t; T pre (i,t) represents the predicted temperature of the i-th power unit at time t; M real (t) represents the actual hotspot migration state at time t; M pre (t) represents the predicted hotspot migration state at time t; u1 and u2 represent the weighting coefficients of the error term; || represents the absolute value operation.

[0105] This formula is used to determine whether the current model needs correction. It compares both the temperature prediction error and the hotspot migration prediction error: if the actual measured temperature deviates significantly from the predicted temperature, or if the actual hotspot migration direction is inconsistent with the prediction result, then E... update An increase in (t) indicates a deviation between the current heat spread map or fan disturbance matrix and the actual operating state. This formula does not directly participate in the control solution; rather, it serves as an online verification trigger condition, determining whether to write back and correct the previously established model parameters, thereby enabling the entire heat dissipation control process to possess continuous adaptive capability.

[0106] Step S54, when the model corrects the trigger amount E update When (t) is greater than the preset correction threshold, the edge weight parameters, propagation priority parameters and region division parameters of the heat diffusion priority propagation map in step S10 are written back and corrected, and the cooling contribution parameters, lateral turbulence parameters and backflow effect parameters of the multi-fan anti-coupling interference matrix in step S20 are corrected synchronously.

[0107] The corrected heat propagation priority map and multi-fan anti-coupling interference matrix are re-output to steps S30 and S40 to form a closed-loop dynamic heat dissipation optimization management process of "heat propagation prediction - interference identification - risk screening - frequency misalignment optimization - result verification - parameter write-back".

[0108] In this embodiment, the present invention provides an intelligent heat dissipation optimization management system for a multi-chip frequency converter driver, including a data acquisition module 101, a heat diffusion construction module 102, an anti-coupling interference matrix module 103, a local thermal risk constraint module 104, a partitioned frequency error optimization module 105, an execution module 106, a real-time monitoring module 107, and an online review and correction module 108.

[0109] The data acquisition module 101 is used to collect the operating status parameters, thermal status parameters and spatial distribution parameters of multiple power units, as well as the installation position, air duct coverage relationship, air supply overlap relationship and feedback operation information of multiple cooling fans.

[0110] The heat diffusion construction module 102 is used to receive the operating status parameters, thermal status parameters and spatial distribution parameters of multiple power units transmitted by the data acquisition module 101, and construct a heat diffusion priority propagation map for hot spot migration prediction by combining the thermal conduction correlation between power units, and dynamically update the heat diffusion priority propagation map.

[0111] The anti-coupling interference matrix module 103 receives the heat diffusion priority propagation map transmitted by the heat diffusion construction module 102, and combines the installation positions of multiple cooling fans, air duct coverage relationship, and air supply overlap relationship to establish a multi-fan anti-coupling interference matrix to describe the positive cooling effect and reverse disturbance effect of multiple cooling fans on different areas, and is used to dynamically update the anti-coupling interference matrix.

[0112] The local thermal risk constraint module 104 is used to receive the thermal diffusion propagation map transmitted by the thermal diffusion construction module 102, receive the anti-coupling interference matrix transmitted by the anti-coupling interference matrix module 103, construct the local thermal risk constraint function, and screen out pseudo-optimal control schemes that can reduce the local target temperature but will cause the adjacent area temperature to deteriorate.

[0113] The partitioned frequency optimization module 105 is used to receive the local thermal risk constraint function transmitted by the local thermal risk constraint module 104, and perform partitioned frequency joint optimization on the target speed, PWM duty cycle and PWM switching frequency of multiple cooling fans based on the local thermal risk constraint function, so as to obtain the optimal fan control parameter combination that satisfies the junction temperature limit of each power unit, reduces the highest hot spot temperature and suppresses the synchronous beat effect of multiple fans.

[0114] Execution module 106 is used to drive multiple cooling fans to operate according to the optimal combination of fan control parameters;

[0115] The real-time monitoring module 107 is used to collect actual temperature distribution data, hot spot migration data and fan feedback data after the execution module is executed;

[0116] The online verification and correction module 108 is used to acquire the actual temperature distribution data, hot spot migration data and fan feedback data transmitted by the real-time monitoring module, and feed the data back to the heat diffusion construction module and the anti-coupling interference matrix module for online verification and parameter write-back correction.

[0117] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent heat dissipation optimization management of a multi-chip frequency converter driver, characterized in that, The method includes the following steps: Step S10: Collect the operating status parameters, thermal status parameters, and spatial distribution parameters of multiple power units, and construct a heat diffusion priority propagation map for hot spot migration prediction by combining the thermal conduction correlation between power units. Step S20: Based on the heat diffusion priority propagation map constructed in step S10, and combined with the installation positions of multiple cooling fans, air duct coverage relationship, and air supply overlap relationship, a multi-fan anti-coupling interference matrix is ​​established to describe the positive cooling effect and reverse disturbance effect of multiple cooling fans on different areas. Step S30: Based on the heat diffusion priority propagation diagram described in step S10 and the multi-fan anti-coupling interference matrix described in step S20, a local thermal risk constraint function is constructed to screen out pseudo-optimal control schemes that can reduce the local target temperature but will cause the adjacent area temperature to deteriorate. Step S40: Under the constraint of the local thermal risk function described in step S30, the target speed, PWM duty cycle and PWM switching frequency of multiple cooling fans are jointly optimized by partitioning and frequency shifting to obtain the optimal fan control parameter combination that satisfies the junction temperature limit of each power unit, reduces the highest hot spot temperature and suppresses the synchronous beat effect of multiple fans. In step S50, multiple cooling fans are driven to run according to the optimal fan control parameter combination obtained in step S40. Based on the actual temperature distribution, hot spot migration results and fan feedback results after control execution, online verification and parameter write-back correction are performed on the heat diffusion priority propagation map in step S10 and the multi-fan anti-coupling interference matrix in step S20.

2. The intelligent heat dissipation optimization management method for a multi-chip frequency converter driver as described in claim 1, characterized in that, Step S10 includes: Step S11: Collect the operating current, operating voltage, switching status, loss estimate, case temperature or near-junction temperature, local ambient temperature and load rate information of multiple power units, and obtain the spatial coordinates, adjacency relationship, common heat dissipation structure relationship and heat conduction path relationship of each power unit inside the driver. Step S12: Calculate the priority weight of the thermal influence propagated from the i-th power unit to the j-th power unit based on the equivalent heating power, temperature rise rate, thermal structure correlation coefficient between the i-th power unit and the j-th power unit, and thermal path length. Step S13: Establish a directed heat propagation relationship from the source power unit to the target power unit according to the priority weight between each power unit, and construct a heat diffusion priority propagation map for hot spot migration prediction based on each directed heat propagation relationship.

3. The intelligent heat dissipation optimization management method for a multi-chip frequency converter driver as described in claim 1, characterized in that, Step S20 includes: Step S21: Based on the heat diffusion priority propagation map obtained in step S10, the regions where multiple power units are located are dynamically divided to obtain the hot spot core area, hot spot diffusion front area, ordinary heating area and reverse cooling sensitive area. Step S22: Obtain the installation position, air outlet direction, rated air volume, current speed, PWM drive parameters, air duct coverage boundary, and relative positional relationship between multiple cooling fans; Step S23: Calculate the overall interference intensity of the m-th cooling fan on the n-th region based on the lateral turbulence intensity, backflow effect intensity, and actual effective cooling contribution generated by the m-th cooling fan on the n-th region. Step S24: Based on the comprehensive interference intensity of each cooling fan in each region, establish the mapping relationship of the effects of multiple cooling fans on different regions, and extract the effective cooling contribution of each cooling fan to the target region, the lateral turbulence influence on adjacent regions, the backflow effect on adjacent air ducts, and the weakening effect on local air supply uniformity, to form a multi-fan anti-coupling interference matrix.

4. The intelligent heat dissipation optimization management method for a multi-chip frequency converter driver as described in claim 1, characterized in that, Step S30 includes: Step S31: Based on the heat diffusion priority propagation map output in step S10, extract the hot spot core region, the hot spot migration front region, and the candidate affected region that has a heat propagation relationship with the hot spot migration front region from multiple power units. Step S32: Based on the multi-fan anti-coupling interference matrix output in step S20, extract the positive cooling component and negative disturbance component of each cooling fan to the hot spot core region, the hot spot migration front region and the candidate affected region. Step S33: Construct a local thermal risk constraint function based on the predicted maximum junction temperature of all power units, the positive temperature rise increment of non-target areas, the comprehensive interference intensity of each cooling fan on different areas, and the correlation of the fan's effect on the corresponding area. Step S34: Based on the local thermal risk constraint function, perform risk screening on the candidate fan control schemes and eliminate pseudo-optimal control schemes that, although they can reduce the local target temperature, may cause reverse heating in adjacent areas, accelerated hot spot migration, or enhanced local disturbances.

5. The intelligent heat dissipation optimization management method for a multi-chip frequency converter driver as described in claim 1, characterized in that, Step S40 includes: Step S41: Based on the evaluation of the local thermal risk constraint function constructed in step S30, the target speed, PWM duty cycle and PWM switching frequency of multiple cooling fans are defined as the control variables to be optimized, and the frequency interval relationship, phase stagger relationship and regional primary and secondary coverage relationship between multiple fans are incorporated into a unified parameter space. Step S42: Based on the hot spot diffusion direction indicated by the heat diffusion priority propagation map described in step S10, prioritize the allocation of fan adjustment authority and adjustment range for the corresponding area, and restrict the synchronous speed-up behavior and same-frequency drive behavior of adjacent fans according to the multi-fan anti-coupling interference matrix described in step S20. Step S43: Jointly evaluate the local thermal risk target value corresponding to the candidate control parameter combination, the total power consumption of all cooling fans, and the synchronous beat intensity index of multiple cooling fans to obtain the comprehensive optimization target value; Step S44: Under the conditions of meeting the upper limit of junction temperature of each power unit, the allowable air loss boundary of the system and the local disturbance tolerance, perform partitioned frequency misalignment joint optimization for multiple cooling fans to obtain the optimal fan control parameter combination that minimizes the comprehensive optimization objective value.

6. The intelligent heat dissipation optimization management method for a multi-chip frequency converter driver as described in claim 1, characterized in that, Step S50 includes: Step S51: Based on the optimal fan control parameter combination output in step S40, send corresponding target speed commands, PWM duty cycle commands and PWM switching frequency commands to multiple cooling fans respectively, so that multiple cooling fans can operate in coordination according to a predetermined partitioned frequency misalignment control method. Step S52: After control execution, continuously collect the actual temperature value, temperature rise rate value, hot spot location change result, hot spot migration path change result, actual speed value of each cooling fan, feedback current value, local wind speed feedback value, and air duct pressure difference feedback value of each power unit. Step S53: Calculate the model correction trigger amount based on the actual temperature, predicted temperature, actual hot spot migration status, and predicted hot spot migration status of the power unit. Step S54: When the model correction trigger amount is greater than the preset correction threshold, the edge weight parameters, propagation priority parameters and region division parameters of the heat diffusion priority propagation map in step S10 are written back and corrected, and the cooling contribution parameters, lateral turbulence parameters and backflow effect parameters of the multi-fan anti-coupling interference matrix in step S20 are corrected synchronously. In step S55, the corrected heat diffusion priority propagation map and multi-fan anti-coupling interference matrix are re-output to steps S30 and S40 to form a closed-loop dynamic heat dissipation optimization management process.

7. The intelligent heat dissipation optimization management system for a multi-chip frequency converter driver as described in any one of claims 1-6, characterized in that, include: The data acquisition module is used to collect the operating status parameters, thermal status parameters, and spatial distribution parameters of multiple power units, as well as the installation position, air duct coverage relationship, air supply overlap relationship, and feedback operation information of multiple cooling fans. The heat diffusion construction module is used to receive the operating status parameters, thermal status parameters and spatial distribution parameters of multiple power units transmitted by the data acquisition module, and construct a heat diffusion priority propagation map for hot spot migration prediction by combining the thermal conduction correlation between power units, and dynamically update the heat diffusion priority propagation map. The anti-coupling interference matrix module receives the heat diffusion priority propagation map transmitted by the heat diffusion construction module, and combines the installation positions of multiple cooling fans, air duct coverage relationship, and air supply overlap relationship to establish a multi-fan anti-coupling interference matrix to describe the positive cooling effect and reverse disturbance effect of multiple cooling fans on different areas, and is used to dynamically update the anti-coupling interference matrix. The local thermal risk constraint module is used to receive the thermal diffusion propagation map transmitted by the thermal diffusion construction module, receive the anti-coupling interference matrix transmitted by the anti-coupling interference matrix module, construct the local thermal risk constraint function, and screen out pseudo-optimal control schemes that can reduce the local target temperature but will cause the adjacent area temperature to deteriorate. The partitioned frequency optimization module is used to receive the local thermal risk constraint function transmitted by the local thermal risk constraint module, and perform partitioned frequency joint optimization on the target speed, PWM duty cycle and PWM switching frequency of multiple cooling fans based on the local thermal risk constraint function, so as to obtain the optimal fan control parameter combination that satisfies the junction temperature limit of each power unit, reduces the highest hot spot temperature and suppresses the synchronous beat effect of multiple fans. The execution module is used to drive multiple cooling fans to operate according to the optimal combination of fan control parameters; The real-time monitoring module is used to collect actual temperature distribution data, hot spot migration data, and fan feedback data after the execution module is executed. The online verification and correction module is used to acquire the actual temperature distribution data, hot spot migration data and fan feedback data transmitted by the real-time monitoring module, and feed the data back to the heat diffusion construction module and the anti-coupling interference matrix module for online verification and parameter write-back correction.

Images