Controller energy consumption-temperature rise integrated optimization method in industrial control scenario

Abstract

This invention relates to the field of temperature control technology and discloses an integrated optimization method for controller energy consumption and temperature rise in industrial control scenarios. A sliding window is used to collect the average load and extreme temperature values, mapping them to generate a three-dimensional heat distribution matrix and filtering heat dissipation bottlenecks. A two-layer three-dimensional mesh thermal coupling model with a heating layer and a heat dissipation layer is constructed, with each mesh configured with heating intensity and heat dissipation capacity. Thermal mismatch nodes are marked using normalization, and virtual heat flow pathways are planned and matched with dynamic response time windows. High-temperature switching events are split and inserted into the response window in a dispersed manner, generating a drive level sequence with dead-zone intervals and sending it to the drive circuit as a temperature rise adjustment execution command. Temperature is collected in a closed-loop cycle and compared with the temperature rise envelope target. Over-temperature conditions result in reduced duty cycle, while the safe range only requires fine-tuning the frequency. The closed-loop system achieves convergent adjustment of device temperature rise. This invention achieves precise matching of spatial heating and heat dissipation resources, balancing temperature regulation accuracy and equipment energy consumption optimization, without requiring additional heat dissipation mechanisms, and is suitable for complex industrial conditions.

Description

Technical Field

[0001] This invention relates to the field of temperature control technology, specifically to an integrated optimization method for controller energy consumption and temperature rise in industrial control scenarios. Background Technology

[0002] During the operation of frequency converters and servo controllers in industrial settings, power devices continuously generate conduction and switching losses. The accumulation of heat from these losses causes the device temperature to rise continuously. Existing mature temperature control solutions only use the average temperature of the entire machine to perform global and uniform temperature control on the devices. This type of temperature control only collects a single temperature value of the entire machine to carry out closed-loop temperature control, and cannot distinguish the intensity of heat generation and heat dissipation capacity of different areas inside the controller.

[0003] When the total heat generated in a local space exceeds the heat dissipation limit of that area, the existing temperature regulation logic can only reduce the overall output capacity of the whole machine to suppress the device temperature. The remaining areas of the whole machine with sufficient heat dissipation margin cannot be fully utilized, resulting in an imbalance in the allocation of heat dissipation resources. At the same time, the global uniform adjustment of the switching action will indiscriminately increase the operating losses of all power devices. In the process of realizing device temperature regulation, it is difficult to take into account both the equipment output level and the loss control requirements. There is a clear contradiction between temperature regulation accuracy and equipment operating economy. Summary of the Invention

[0004] The purpose of this invention is to provide an integrated optimization method for controller energy consumption and temperature rise in industrial control scenarios, in order to solve two sets of technical contradictions in the temperature control of controllers in the prior art: First, local high-heat areas need to be cooled down and suppressed, while the heat dissipation margin of other areas is sufficient but not fully utilized, making it difficult to balance temperature suppression and equipment output maintenance; Second, when the temperature exceeds the standard, it needs to be cooled down quickly, while no additional losses should be added within the safe range, making it difficult to balance temperature control accuracy and operating economy.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an integrated optimization method for controller energy consumption and temperature rise in industrial control scenarios, used for closed-loop non-electrophysical quantity regulation of the temperature rise of power devices inside industrial controllers, including: S1. Synchronously collect real-time load parameters and multi-point temperature parameters from the controller, bind the timestamp to generate parameter pair sequences, and output the load mean and temperature extreme value statistical features through a sliding window buffer. Only the two types of statistical features are used as the sole input for subsequent calorific value matrix and temperature field mapping modeling, discarding the instantaneous original parameters. S2. Map statistical characteristics to the power module loss map, project load along the loss trajectory to generate a three-dimensional heat generation distribution matrix; simultaneously fit the global temperature gradient field, screen out abnormal gradient regions at the heat dissipation channel outlets as heat dissipation bottlenecks, and then... Calculate the characteristic value of the bottleneck thermal resistance, where, The characteristic value of thermal resistance in the heat dissipation bottleneck area, in °C / W; The steady-state temperature difference between the two ends of the heat dissipation bottleneck area, in °C. The steady-state heat flux through the heat dissipation bottleneck region, in W; S3. Establish a three-dimensional mesh coordinate system matching the physical topology of the controller, construct heat-generating and heat-dissipating layer meshes layer by layer, calculate the interlayer thermal coupling coefficient by combining mesh spacing and medium thermal conductivity, and map heat generation intensity and thermal resistance to the corresponding meshes to construct a thermal-energy coupling model. Configure heat generation intensity and heat dissipation capacity fields for each mesh. ,in, The heat dissipation capacity of the grid represents the maximum heat dissipation capacity that the grid can withstand, expressed in W / ℃. The characteristic value of thermal resistance in the heat dissipation bottleneck area, in °C / W; The reference thermal conductivity area corresponding to the grid, in meters. 2 ; S4. Normalize the grid heat intensity and heat dissipation capacity to the global maximum value, and mark the grid with normalized heat intensity greater than heat dissipation capacity as thermal mismatch nodes; search for surrounding redundant heat dissipation grids by Manhattan distance, plan the shortest virtual heat flow conduction path without negative capacity, and adaptively match the corresponding time window according to the path thermal resistance and thermal capacity response time and thermal mismatch over-limit amplitude, and encapsulate the path and window parameters into dynamic thermal constraints. S5. Match the switching sequence power devices and thermal mismatch module, mark the corresponding high-temperature switching events; split the heating pulses equally according to the number of windows and disperse them into each response window to generate a recombined switching sequence; convert it into a drive level sequence with device dead zone interval as a temperature control command; S6. Send temperature control commands to the drive circuit, using the target temperature setpoint of the power device as the control target for non-electrical physical quantity adjustment, periodically collect the actual temperature and compare it with the target temperature setpoint; if the actual temperature is too high, reduce the switching duty cycle to reduce heat generation; if the actual temperature is too low, maintain the switching duty cycle unchanged and only finely adjust the switching frequency; through closed-loop feedback adjustment, the actual temperature converges to the target temperature setpoint.

[0006] As a preferred embodiment of the present invention, the specific steps for generating the three-dimensional heat distribution matrix in step S2 are as follows: Step 1: Extract two types of load values: real-time on-state current and switching frequency of a single power module; Step 2: Project the on-state current along the on-state loss trajectory to the module terminals, solder joints, and internal cable mesh, and accumulate them to obtain the on-state heating base value; Step 3: Project the switching frequency along the switching loss trajectory onto the module driver gate and parasitic inductance grid, and sum them to obtain the base value of switching heating. Step 4: Overlay the on-state and switching heating base values ​​within the same 3D mesh, and fill the corresponding coordinates of the heat distribution matrix with the overlay result as the mesh heating intensity.

[0007] As a preferred embodiment of the present invention, the specific steps for S2 to extract the heat dissipation bottleneck area are as follows: Step 1: Read the coordinates of all temperature measurement points and the measured temperature values ​​of the casing and power devices in the temperature field distribution diagram; Step 2: Based on the spatial location and temperature value of each measuring point, calculate the magnitude of the temperature gradient vector grid by grid. Step 3: Define the grid with a gradient modulus greater than the preset gradient threshold of 1.5℃ / cm as the temperature gradient anomaly zone; Step 4: Select the mesh set located at the heat sink airflow outlet and the edge of the thermally conductive substrate within the abnormal area as the heat dissipation bottleneck, record the bottleneck boundary coordinates and substitute them into the equation. Calculate the thermal resistance.

[0008] As a preferred embodiment of the present invention, the specific steps for constructing the thermal-energy coupling model in step S3 are as follows: Step 1: Divide the 3D mesh into X / Y axes to match the length and width of the controller, and divide the Z axis into power device heat dissipation layer and housing heat dissipation layer; Step 2: Fill the heating intensity of all grids in the heat generation distribution matrix into the heating layer coordinate by coordinate; map the heat dissipation bottleneck thermal resistance to the corresponding grid of the heat dissipation layer according to the boundary coordinates; Step 3: Take the straight-line distance between the centers of the vertically aligned heat-generating and heat-dissipating grids, and the preset thermal conductivity of the heat-conducting medium between the grids is 2.5W / (m·K), and calculate the interlayer thermal coupling coefficient; Step 4: Use the thermal coupling coefficient to associate the upper and lower layers of meshes with the same coordinates to complete the binding of the two-layer meshes and form a thermal-energy coupled association model.

[0009] As a preferred embodiment of the present invention, the specific operation steps for S4 to identify thermal mismatch nodes are as follows: Step 1: Traverse all 3D meshes of the model and retrieve the original heat intensity and original heat dissipation capacity values ​​for each cell; Step 2: Extract the maximum heat intensity and maximum heat dissipation capacity of the entire mesh as normalization benchmarks; Step 3: Calculate the dimensionless comparison values ​​respectively: Normalized heat generation value = heat generation intensity of a single cell / maximum heat generation intensity of the entire area, Normalized heat dissipation value = heat dissipation capacity of a single cell / maximum heat dissipation capacity of the entire area; Step 4: Compare the two sets of dimensionless values. If the normalized calorific value is larger, mark the current grid as a thermal mismatch node, indicating an exceedance of the limit. ,in, The thermal mismatch node's overheating exceeds the limit, dimensionless, representing the relative proportion of mesh heating exceeding heat dissipation capacity; The original heat intensity of a single grid, in W. This represents the original heat dissipation capacity of a single grid, in W / ℃.

[0010] As a preferred embodiment of the present invention, the specific operation steps of S4 planning the virtual heat flow conduction path and response window are as follows: Step 1: Using the thermal mismatch mesh as the center, search all surrounding meshes within a preset maximum radius of 8 squares from near to far according to the Manhattan distance; Step 2: Select grids that simultaneously meet the conditions of normalized heat dissipation value > normalized heat generation value and grid heat dissipation capacity > 0 as redundant heat dissipation nodes; Step 3: Connect the grid edges of redundant nodes to mismatched nodes to generate multiple candidate path chains, and eliminate invalid paths that pass through grids with heat dissipation capacity ≤ 0; Step 4: Select the candidate path with the fewest grid nodes as the optimal virtual heat flow conduction path; Step 5: Substitute the path equivalent thermal resistance and equivalent heat capacity to solve for the first-order system step response time. Combine the thermal mismatch over-limit amplitude adaptive correction time and set the corrected time as the path-specific response time window length.

[0011] As a preferred embodiment of the present invention, the specific operation steps for the S5 splitting of the high-temperature switch event are as follows: Step 1: Read the original switch sequence one by one, and record the trigger time and the bound power device number of each switch event; Step 2: Match the power device number with the power module to which the thermal mismatch node output by S4 belongs. If the match is successful, mark the event as a high thermal switching event. Step 3: Extract the total heating pulse area and total duration of the high-thermal-switching event, and read the total number N of response windows within the dynamic thermal constraint; Step 4: Divide the original heating pulse area into N equal parts to generate N sub-switch events, with each sub-switch allocated the same pulse area; Step 5: Insert the N sub-switch events into the N independent response time windows at uniform time intervals, and output the recombined switch sequence.

[0012] As a preferred embodiment of the present invention, the specific steps for generating the temperature control command in step S5 are as follows: Step 1: Traverse all sub-switch events in the recombination switch sequence and generate corresponding drive pin level transition signals. The rising edge is the device turn-on time, the falling edge is the turn-off time, and the interval between the rising and falling edges is equal to the sub-switch conduction time. Step 2: Extract the measured turn-on delay and turn-off delay of the power device, and take the larger of the two values ​​as the preset device dead time reference of 0.8μs; Step 3: Insert a 0.8μs dead interval between two adjacent sets of level transition signals; Step 4: Connect all the level transition signals in sequence according to time to form a continuous drive level sequence, which serves as the temperature control command sent to the drive circuit to regulate the temperature rise of the device.

[0013] As a preferred embodiment of the present invention, the specific operation steps of the S6 closed-loop temperature control convergence regulation are as follows: Step 1: Collect the real-time temperature of the power devices inside the controller every 100ms preset acquisition period, and read the preset target temperature of the temperature rise envelope at the current moment; Step 2: Perform a difference calculation to obtain the temperature deviation. A deviation > 0 indicates that the actual temperature exceeds the standard. Step 3: When the deviation is greater than 0, output a duty cycle reduction command to the drive circuit to synchronously reduce the proportion of single-switch cycle conduction time according to the deviation ratio. Step 4: When the deviation is ≤0, keep the current duty cycle unchanged, and only slightly adjust the switching frequency within the ±1kHz preset frequency adjustment range to improve the dynamic response of the device; Step 5: Repeatedly execute the data acquisition, comparison, and adjustment process to gradually reduce the absolute value of the deviation between the actual temperature and the target temperature until stable convergence.

[0014] As a preferred embodiment of the present invention, the specific operation steps of the S1 sliding window preprocessing parameters on the sequence are as follows: Step 1: Input all load and temperature parameters with uniform timestamps into a fixed-width sliding window buffer; Step 2: The cache executes the first-in, first-out rule, and when a new parameter pair enters, it automatically removes the set of parameters that has been stored in the window the longest. Step 3: After the window is filled, calculate the arithmetic mean of all load parameters and the maximum value of all temperature parameters within the window in real time. Step 4: Output the average load and maximum temperature as the only input data to the S2 loss map and temperature field mapping process.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention addresses the closed-loop regulation of device temperature rise, a non-electrical physical quantity. It incorporates a multi-source data smoothing preprocessing step before the temperature regulation process, and relies on window statistical features to participate in the global thermal field modeling. This avoids interference from instantaneous load and instantaneous temperature fluctuations on the temperature regulation model, ensuring that the temperature regulation judgment is aligned with the long-term stable operating conditions of the equipment, thereby improving the regulation stability of the entire temperature rise closed-loop regulation system.

[0016] 2. This invention constructs a hierarchical grid thermal coupling model to synchronously quantify the heat generation capacity and heat dissipation capacity limit of various locations in space. By relying on dimensionless conversion to unify the comparison standard of the two types of physical quantities, it can accurately locate the area inside the device where local heat generation exceeds the heat dissipation capacity. By relying on the surrounding idle heat dissipation resources to plan virtual thermal constraints, it can achieve fine matching of heat generation and heat dissipation resources in the spatial dimension. It can complete the device temperature suppression and adjustment without reducing the overall device output, and fully release the normal output range of the device.

[0017] 3. This invention performs time-domain splitting and decentralized processing of the switching action corresponding to local high heat. The switching timing adjustment is only used as an execution means to achieve device temperature regulation. It only adjusts the instantaneous heat flow output in the area of ​​heat imbalance, while the other power devices maintain their original working timing. Unlike the traditional temperature control logic of global unified parameter adjustment, it can effectively reduce the temperature impact caused by short-term concentrated heat generation. The temperature regulation process will not indiscriminately limit the overall machine performance.

[0018] 4. This invention sets up a closed-loop temperature regulation logic that differentiates between different operating conditions. It reduces the heating power only when the device temperature exceeds the preset range, and does not actively increase additional operating losses when the device temperature is within the safe range. While continuously keeping the device temperature within the preset temperature rise curve, it simultaneously reduces the energy loss generated by the long-term operation of the equipment, thus taking into account both the temperature regulation accuracy and the energy consumption control of the equipment.

[0019] 5. The entire temperature regulation process forms a complete self-consistent temperature rise regulation and control link, from data preprocessing, spatial thermal field modeling, local thermal balance constraints to time-domain switch scheduling and closed-loop temperature correction. All regulation steps are completed autonomously based on the load and temperature data collected by the equipment itself, without the need for additional heat dissipation actuators or manual intervention. It is suitable for various complex working conditions of continuous operation and frequent start-stop in various industries, and the temperature regulation system has stronger environmental adaptability. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the hierarchical structure of the three-dimensional mesh thermal-energy coupling model in the integrated optimization method for controller energy consumption and temperature rise in industrial control scenarios of the present invention. Detailed Implementation

[0021] 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, and 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. Example 1

[0022] like Figure 1As shown, this invention provides an integrated optimization method for controller energy consumption and temperature rise in industrial control scenarios, used to achieve closed-loop regulation of non-electrical physical quantities of temperature rise of power devices inside the controller, including: S1. Synchronously collect real-time load parameters and multi-point temperature parameters from the controller, bind the timestamp to generate parameter pair sequences, and output the load mean and temperature extreme value statistical features through a sliding window buffer. Only the two types of statistical features are used as the sole input for subsequent calorific value matrix and temperature field mapping modeling, discarding the instantaneous original parameters. Before this step is executed, multiple synchronous acquisition channels are pre-deployed, with the sampling period of each channel uniformly set to 100ms. All acquired data are synchronously marked with a uniform numerical timestamp to eliminate modeling errors caused by misalignment of acquisition timing for different physical quantities. The sliding window buffer is configured with a fixed window length, which can accommodate two hundred sets of continuous parameter pairs. The buffered data strictly follows the first-in-first-out storage and update rule. When a new set of load and temperature parameter pairs is stored in the buffer, the parameter pair with the longest storage time is automatically removed. After the window is filled with data, two types of fixed statistical features are continuously output: one is the arithmetic mean of all real-time load parameters within the window, and the other is the maximum value of all multi-point temperature parameters within the window. The instantaneous acquisition values ​​at a single moment are not directly used in subsequent modeling calculations to avoid interference with the accuracy of thermal field modeling caused by instantaneous load impacts and instantaneous temperature fluctuations.

[0023] S2. Map statistical characteristics to the power module loss map, project load along the loss trajectory to generate a three-dimensional heat generation distribution matrix; simultaneously fit the global temperature gradient field, screen out abnormal gradient regions at the heat dissipation channel outlets as heat dissipation bottlenecks, and then... Calculate the characteristic value of the bottleneck thermal resistance, where, The characteristic value of thermal resistance in the heat dissipation bottleneck area, in °C / W; The steady-state temperature difference between the two ends of the heat dissipation bottleneck area, in °C. The steady-state heat flux through the heat dissipation bottleneck region, in W; The loss map is pre-stored in the controller's local storage unit. The map distinguishes between two types of fixed loss projection paths: the on-state loss projection path covers the power device's lead-out terminals, solder contact surfaces, and internal conductive copper busbars; the switching loss projection path covers the power device's drive pins, gate metal layer, and parasitic inductance region. The average output load value of the window is projected along these two paths to the corresponding coordinates of the 3D grid. The heat generation base values ​​obtained from both projections are superimposed at the same grid location, and the superimposed values ​​are directly filled into the corresponding grid of the 3D heat generation distribution matrix. Based on the spatial coordinates of all casing and power device temperature measurement points and the collected temperature values, the temperature gradient vector is solved grid by grid. Grids with a vector magnitude greater than 1.5℃ per centimeter are uniformly designated as abnormal temperature gradient regions. Within all abnormal grids, a continuous set of grids located at the radiator air duct outlet is selected as the heat dissipation bottleneck region. The stable temperature difference between the two ends of the bottleneck region is read. With steady-state heat flow through the region Substituting into a fixed formula, the unique thermal resistance characteristic value of this region is obtained. .

[0024] S3. Establish a three-dimensional mesh coordinate system matching the physical topology of the controller, construct heat-generating and heat-dissipating layer meshes layer by layer, calculate the interlayer thermal coupling coefficient by combining mesh spacing and medium thermal conductivity, and map heat generation intensity and thermal resistance to the corresponding meshes to construct a thermal-energy coupling model. Configure heat generation intensity and heat dissipation capacity fields for each mesh. ,in, The heat dissipation capacity of the grid represents the maximum heat dissipation capacity that the grid can withstand, expressed in W / ℃. The characteristic value of thermal resistance in the heat dissipation bottleneck area, in °C / W; The reference thermal conductivity area corresponding to the grid, in meters. 2 ; The three-dimensional mesh coordinate system perfectly replicates the overall dimensions of the controller. The vertical Z-axis is divided into two fixed mesh layers: the upper layer represents the power device's heating element, and the lower layer represents the casing's heat dissipation layer. The coordinates of the two mesh layers correspond perfectly. The straight-line distance between the mesh centers serves as the benchmark for calculating inter-layer distances. A thermally conductive medium with a fixed thermal conductivity of 2.5 W / m Kelvin is filled between the mesh layers. The thermal coupling coefficient between the two perpendicular mesh layers is calculated based on the distance and thermal conductivity values. The heating intensity values ​​of all meshes in the three-dimensional heat generation distribution matrix are completely filled into the corresponding coordinates of the heating layer. The thermal resistance characteristic values ​​of the heat dissipation bottleneck area are completely mapped to the corresponding mesh of the heat dissipation layer according to the boundary coordinates. Each set of vertically aligned meshes establishes a numerical relationship based on the calculated thermal coupling coefficient. Each three-dimensional mesh is simultaneously configured with a heating intensity field and a heat dissipation capacity field, with the heat dissipation capacity based on the mesh's reference area. Corresponding thermal resistance characteristic value The calculation is performed according to a fixed formula.

[0025] S4. Normalize the grid heat intensity and heat dissipation capacity to the global maximum value, and mark the grid with normalized heat intensity greater than heat dissipation capacity as thermal mismatch nodes; search for surrounding redundant heat dissipation grids by Manhattan distance, plan the shortest virtual heat flow conduction path without negative capacity, and adaptively match the corresponding time window according to the path thermal resistance and thermal capacity response time and thermal mismatch over-limit amplitude, and encapsulate the path and window parameters into dynamic thermal constraints. Traverse all grids within the 3D mesh coordinate system, extracting the original heat intensity and original heat dissipation capacity values ​​for each cell. After traversal, extract the maximum heat intensity and maximum heat dissipation capacity within all grids as a unified normalization benchmark. The normalized heat value of a single cell is equal to the original heat intensity of that cell divided by the maximum heat intensity of the entire domain, and the normalized heat dissipation value of a single cell is equal to the original heat dissipation capacity of that cell divided by the maximum heat dissipation capacity of the entire domain. Compare the two sets of dimensionless values ​​cell by cell. Grids with larger normalized heat values ​​are identified as thermal mismatch nodes, and the over-limit amplitude is solved simultaneously. Using the thermal mismatch mesh as the search center, all surrounding meshes within an eight-grid radius are searched according to the Manhattan distance from near to far. Meshes that simultaneously meet the conditions of normalized heat dissipation value being greater than normalized heat generation value and mesh heat dissipation capacity being greater than zero are identified as redundant heat dissipation nodes. Multiple candidate path chains are constructed from the redundant heat dissipation nodes to the thermal mismatch nodes along the mesh edges. Invalid paths with heat dissipation capacity less than or equal to zero in any mesh within the path are eliminated, and the path containing the fewest meshes is selected as the optimal virtual heat flow conduction path. The equivalent thermal resistance and equivalent heat capacity of the path are substituted to solve for the basic duration of the first-order system step response. The basic duration is adaptively corrected in conjunction with the thermal mismatch over-limit amplitude. The corrected duration serves as the exclusive response time window for this path. The complete path parameters and window duration parameters are packaged together to form dynamic thermal constraint conditions.

[0026] S5. Match the switching sequence power devices and thermal mismatch module, mark the corresponding high-temperature switching events; split the heating pulses equally according to the number of windows and disperse them into each response window to generate a recombined switching sequence; convert it into a drive level sequence with device dead zone interval as a temperature control command; Read the original operating switch sequence of the controller one by one, and record the trigger time and the unique number of the bound power device for each switch event. Establish a one-to-one matching relationship between the power device number and the power module to which the thermal mismatch node identified in step S4 belongs. The switch events that are successfully matched are uniformly marked as high-heat switch events. Extract the total area of ​​the complete heating pulse and the total duration of the event of the high-heat switch event, and read the total number N of the response time windows inside the dynamic thermal constraint. Divide the total area of ​​the original heating pulse into N parts on average. Each part of the pulse area corresponds to a new sub-switch event. Fill all the sub-switch events into N independent response time windows at uniform time intervals to complete the splitting and recombination to obtain a new recombined switch sequence. Convert all the sub-switch events inside the recombined switch sequence into drive pin level transition signals. The rising edge of the level corresponds to the device turn-on time, and the falling edge of the level corresponds to the device turn-off time. A fixed dead interval is uniformly inserted between two adjacent sets of level transition signals. All level signals are connected in series in chronological order to form a continuous drive level sequence. This sequence is directly used as the temperature control command sent to the drive circuit.

[0027] S6. Send temperature control commands to the drive circuit, using the target temperature setpoint of the power device as the control target for non-electrical physical quantity adjustment, periodically collect the actual temperature and compare it with the target temperature setpoint; if the actual temperature is too high, reduce the switching duty cycle to reduce heat generation; if the actual temperature is too low, maintain the switching duty cycle unchanged and only finely adjust the switching frequency; through closed-loop feedback adjustment, the actual temperature converges to the target temperature setpoint.

[0028] The generated temperature control command is completely sent to the controller's built-in drive circuit. The drive circuit adjusts the on / off state of the power device according to the level sequence. The real-time temperature of the power device is re-acquired every 100ms. The target temperature value at the corresponding moment of the preset temperature rise envelope curve is retrieved, and the difference between the actual temperature and the target temperature is calculated to obtain the real-time temperature deviation. When the temperature deviation is greater than zero, it means that the current device temperature exceeds the preset limit. A duty cycle reduction command is sent to the drive circuit to reduce the proportion of device conduction time within a single switching cycle according to the deviation value. When the temperature deviation is less than or equal to zero, it means that the current device temperature is in the safe range. The current duty cycle is kept constant throughout the process. Only a small adjustment of the switching frequency is allowed in the ±1 kilohertz range to improve the dynamic response performance of the whole machine. The entire closed-loop process of temperature acquisition, deviation calculation, and command issuance is executed in a loop. The absolute value of the difference between the actual temperature and the target temperature is reduced cycle by cycle. The process is continuously iterated until the difference stabilizes and no longer changes, thus completing the convergence control of the temperature rise curve.

[0029] Furthermore, the specific steps for generating the three-dimensional heat distribution matrix in S2 are as follows: Step 1: Extract two types of load values: real-time on-state current and switching frequency of a single power module; Step 2: Project the on-state current along the on-state loss trajectory to the module terminals, solder joints, and internal cable mesh, and accumulate them to obtain the on-state heating base value; Step 3: Project the switching frequency along the switching loss trajectory onto the module driver gate and parasitic inductance grid, and sum them to obtain the base value of switching heating. Step 4: Overlay the on-state and switching heating base values ​​within the same 3D mesh, and fill the corresponding coordinates of the heat distribution matrix with the overlay result as the mesh heating intensity.

[0030] Furthermore, the specific steps for S2 to extract the heat dissipation bottleneck area are as follows: Step 1: Read the coordinates of all temperature measurement points and the measured temperature values ​​of the casing and power devices in the temperature field distribution diagram; Step 2: Based on the spatial location and temperature value of each measuring point, calculate the magnitude of the temperature gradient vector grid by grid. Step 3: Define the grid with a gradient modulus greater than the preset gradient threshold of 1.5℃ / cm as the temperature gradient anomaly zone; Step 4: Select the mesh set located at the heat sink airflow outlet and the edge of the thermally conductive substrate within the abnormal area as the heat dissipation bottleneck, record the bottleneck boundary coordinates and substitute them into the equation. Calculate the thermal resistance.

[0031] Furthermore, the specific steps for constructing the thermal-energy coupling model in S3 are as follows: Step 1: Divide the 3D mesh into X / Y axes to match the length and width of the controller, and divide the Z axis into power device heat dissipation layer and housing heat dissipation layer; Step 2: Fill the heating intensity of all grids in the heat generation distribution matrix into the heating layer coordinate by coordinate; map the heat dissipation bottleneck thermal resistance to the corresponding grid of the heat dissipation layer according to the boundary coordinates; Step 3: Take the straight-line distance between the centers of the vertically aligned heat-generating and heat-dissipating grids, and the preset thermal conductivity of the heat-conducting medium between the grids is 2.5W / (m·K), and calculate the interlayer thermal coupling coefficient; Step 4: Use the thermal coupling coefficient to associate the upper and lower layers of meshes with the same coordinates to complete the binding of the two-layer meshes and form a thermal-energy coupled association model.

[0032] Furthermore, the specific steps for S4 to identify thermal mismatch nodes are as follows: Step 1: Traverse all 3D meshes of the model and retrieve the original heat intensity and original heat dissipation capacity values ​​for each cell; Step 2: Extract the maximum heat intensity and maximum heat dissipation capacity of the entire mesh as normalization benchmarks; Step 3: Calculate the dimensionless comparison values ​​respectively: Normalized heat generation value = heat generation intensity of a single cell / maximum heat generation intensity of the entire area, Normalized heat dissipation value = heat dissipation capacity of a single cell / maximum heat dissipation capacity of the entire area; Step 4: Compare the two sets of dimensionless values. If the normalized calorific value is larger, mark the current grid as a thermal mismatch node, indicating an exceedance of the limit. ,in, The thermal mismatch node's overheating exceeds the limit, dimensionless, representing the relative proportion of mesh heating exceeding heat dissipation capacity; The original heat intensity of a single grid, in W. This represents the original heat dissipation capacity of a single grid, in W / ℃.

[0033] Furthermore, the specific steps for S4 to plan virtual heat flow pathways and response windows are as follows: Step 1: Using the thermal mismatch mesh as the center, search all surrounding meshes within a preset maximum radius of 8 squares from near to far according to the Manhattan distance; Step 2: Select grids that simultaneously meet the conditions of normalized heat dissipation value > normalized heat generation value and grid heat dissipation capacity > 0 as redundant heat dissipation nodes; Step 3: Connect the grid edges of redundant nodes to mismatched nodes to generate multiple candidate path chains, and eliminate invalid paths that pass through grids with heat dissipation capacity ≤ 0; Step 4: Select the candidate path with the fewest grid nodes as the optimal virtual heat flow conduction path; Step 5: Substitute the path equivalent thermal resistance and equivalent heat capacity to solve for the first-order system step response time. Combine the thermal mismatch over-limit amplitude adaptive correction time and set the corrected time as the path-specific response time window length.

[0034] Furthermore, the specific operational steps for handling the S5 high-temperature switch event are as follows: Step 1: Read the original switch sequence one by one, and record the trigger time and the bound power device number of each switch event; Step 2: Match the power device number with the power module to which the thermal mismatch node output by S4 belongs. If the match is successful, mark the event as a high thermal switching event. Step 3: Extract the total heating pulse area and total duration of the high-thermal-switching event, and read the total number N of response windows within the dynamic thermal constraint; Step 4: Divide the original heating pulse area into N equal parts to generate N sub-switch events, with each sub-switch allocated the same pulse area; Step 5: Insert the N sub-switch events into the N independent response time windows at uniform time intervals, and output the recombined switch sequence.

[0035] Furthermore, the specific steps for S5 to generate temperature control commands are as follows: Step 1: Traverse all sub-switch events in the recombination switch sequence and generate corresponding drive pin level transition signals. The rising edge is the device turn-on time, the falling edge is the turn-off time, and the interval between the rising and falling edges is equal to the sub-switch conduction time. Step 2: Extract the measured turn-on delay and turn-off delay of the power device, and take the larger of the two values ​​as the preset device dead time reference of 0.8μs; Step 3: Insert a 0.8μs dead interval between two adjacent sets of level transition signals; Step 4: Connect all the level transition signals in sequence according to time to form a continuous drive level sequence as the temperature control command sent to the drive circuit.

[0036] Furthermore, the specific operating steps for the S6 closed-loop temperature control convergence regulation are as follows: Step 1: Collect the real-time temperature of the power devices inside the controller every 100ms preset acquisition period, and read the preset target temperature of the temperature rise envelope at the current moment; Step 2: Perform a difference calculation to obtain the temperature deviation. A deviation > 0 indicates that the actual temperature exceeds the standard. Step 3: When the deviation is greater than 0, output a duty cycle reduction command to the drive circuit to synchronously reduce the proportion of single-switch cycle conduction time according to the deviation ratio. Step 4: When the deviation is ≤0, keep the current duty cycle unchanged, and only slightly adjust the switching frequency within the ±1kHz preset frequency adjustment range to improve the dynamic response of the device; Step 5: Repeatedly execute the data acquisition, comparison, and adjustment process to gradually reduce the absolute value of the deviation between the actual temperature and the target temperature until stable convergence.

[0037] Furthermore, the specific operation steps of the S1 sliding window preprocessing parameters on the sequence are as follows: Step 1: Input all load and temperature parameters with uniform timestamps into a fixed-width sliding window buffer; Step 2: The cache executes the first-in, first-out rule, and when a new parameter pair enters, it automatically removes the set of parameters that has been stored in the window the longest. Step 3: After the window is filled, calculate the arithmetic mean of all load parameters and the maximum value of all temperature parameters within the window in real time. Step 4: Output the average load and maximum temperature as the only input data to the S2 loss map and temperature field mapping process. Example 2

[0038] This embodiment builds a logical deduction and verification process based on all the technical features of embodiment 1, and is adapted to the application scenario of industrial robot with frequent start and stop of 11kW servo controller. The servo controller is equipped with four groups of IPM power devices, and eight NTC temperature acquisition points are set inside the whole machine. The narrow and long air duct layout is prone to forming local heat dissipation bottlenecks.

[0039] The first layer of logic is a unified preprocessing logic for multi-source data, corresponding to the complete technical features of S1. The refined technical means include synchronously acquiring load and temperature parameters through multiple acquisition channels, uniformly marking timestamps to form load and temperature parameter pairs, storing these pairs in a fixed-length sliding window buffer for first-in-first-out updates, and stably outputting two statistical features: the average load value and the maximum temperature value. Instantaneous acquired values ​​are discarded throughout the modeling calculations. This layer of logic derives the original technical effect: synchronous timestamp binding eliminates the timing misalignment of load and temperature acquisition; the sliding window's mean and extreme value statistics smooth out data distortion caused by short-term load impacts and local instantaneous temperature spikes. The input data provided for subsequent thermal field modeling has long-term operating condition representativeness, avoiding deviations in determining heat intensity and heat dissipation capacity caused by modeling based on single-moment instantaneous data, thus improving the overall matching accuracy of the system's thermal field model from the data source.

[0040] The second layer of logic is the 3D mesh thermal-energy coupling modeling and thermal mismatch localization logic, corresponding to the complete technical features of S2, S3, and S4. The detailed technical means are divided into three stages of execution. The first stage relies on the statistical features output by the window to solve the global 3D heat generation distribution matrix and the global temperature gradient field, screens the heat dissipation bottleneck area at the air duct outlet, and solves the regional thermal resistance characteristic value. The second stage establishes a dual-layer 3D mesh coordinate system. The heating layer stores the heating intensity of all meshes, while the heat dissipation layer stores the corresponding thermal resistance characteristic values ​​of the meshes. The interlayer thermal coupling coefficient is calculated based on the mesh spacing and the thermal conductivity of the medium to complete the numerical association between the two meshes. Each mesh cell is simultaneously configured with a heating intensity field and a heat dissipation capacity field, with the heat dissipation capacity determined by the formula... The solution is completed; in the third stage, the global maximum value normalization operation is performed on the heat intensity and heat dissipation capacity of all grids, and the dimensionless numerical markers of thermal mismatch nodes are compared. The surrounding redundant heat dissipation grids are searched according to the Manhattan distance with the thermal mismatch nodes as the center, the optimal virtual heat flow conduction path is selected, and the exclusive response time window is adaptively solved by combining the path thermal resistance and thermal capacity and the thermal mismatch over-limit amplitude. The path and window parameters are packaged to form dynamic thermal constraint conditions.

[0041] This layer of logic, derived step-by-step, yields multi-level native technical effects. Segmenting and differentiating between on-state and switching loss projection paths allows for precise location of every subdivided heat source within the controller. Utilizing temperature gradient vectors to filter heat dissipation bottlenecks at the air duct outlet accurately pinpoints inherently weak areas in the overall system's heat dissipation. A dual-layer mesh coupling model synchronously correlates the heat dissipation capacity and heat capacity limits at the same spatial location. Normalized dimensionless conversion resolves the inherent industry problem of incomparable dimensions between heat dissipation power and heat dissipation capacity. Manhattan distance-based retrieval of redundant heat dissipation meshes prioritizes idle heat dissipation resources closest to the thermal mismatch node with sufficient heat dissipation margin. The shortest path length filtering rule reduces the equivalent thermal resistance of heat flow transfer. The response time window adaptively adjusts with the thermal mismatch exceedance; the greater the heat dissipation exceedance, the longer the total window duration and the higher the number of pulse segments, resulting in a synchronously enhanced temporal dispersion of more severe localized heating. The entire modeling and positioning logic achieves refined thermal balance matching across the entire system's spatial dimensions, accurately identifying idle heat dissipation resources and preventing issues like localized overload and unbalanced idle resources.

[0042] The third layer of logic is the high-thermal switching event time-domain splitting and recombination logic, corresponding to the complete technical features of S5. The refined technical means is to match the power device number of the original switching sequence with the corresponding power module for thermal mismatch, mark the successfully matched high-thermal switching events, and split the high-thermal pulse equally according to the total number of response time windows within the dynamic thermal constraint. The sub-switching events are evenly distributed and inserted into the corresponding time-domain windows. The recombined sequence is converted into a drive level sequence with a fixed dead interval as a temperature control command. This layer of logic derivation yields the original technical effect. Traditional controller temperature control methods can only globally and uniformly reduce the switching frequency of the entire machine, and all power devices synchronously reduce the output performance. This layer of logic only splits the switching pulses of the corresponding power devices for the identified thermal mismatch in the time domain. The switching sequence of other power devices without thermal mismatch remains unchanged. It only performs instantaneous heat flux clipping on local overheated areas and does not indiscriminately limit the output capacity of all power devices in the whole machine. The equal splitting of high heat pulses and the uniform time domain dispersion can significantly reduce the peak value of instantaneous heat flux density in a single period, eliminate the temperature step surge caused by short-term high-power switching action, and the dead interval is uniformly set according to the maximum delay of the device. The splitting and recombining of the timing will not cause the risk of bridge arm shoot-through short circuit, and ensures the electrical operation safety of the controller while suppressing the instantaneous temperature rise impact.

[0043] The fourth layer of logic is a closed-loop temperature convergence control logic based on different operating conditions. It corresponds to the complete technical features of S6 and refines the technical means by periodically collecting the real-time temperature of the device, comparing it with the preset temperature rise envelope target temperature to solve for the temperature deviation. When the deviation is greater than zero, the duty cycle of the power device is reduced to decrease the real-time heat generation power. When the deviation is less than or equal to zero, the duty cycle is kept constant, and the switching frequency is only slightly adjusted within the ±1 kHz range to optimize dynamic performance. This continuous closed-loop iteration reduces the absolute value of the temperature deviation until stable convergence. This layer of logic derives the original technical effect, surpassing traditional temperature control methods. When the device temperature is below the target limit, the system will actively increase the duty cycle to increase heat loss and artificially generate excess power consumption. This layer of logic distinguishes between two types of operating conditions and controls them differently. In the case of device over-temperature, the system directly reduces the conduction time to reduce real-time heat power from the source and quickly lowers the device temperature back to the preset envelope range. In the case of device low-temperature safety, no additional heat loss is added. Only the switching frequency is slightly adjusted to take into account the dynamic response requirements of the device. While achieving precise convergence of temperature rise, the system continuously reduces the average switching loss of the whole machine, taking into account both the temperature control accuracy and the optimization of the whole machine's energy consumption.

[0044] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

Claims

1. An integrated optimization method for controller energy consumption and temperature rise in industrial control scenarios, used for closed-loop non-electrophysical quantity regulation of the temperature rise of power devices inside industrial controllers, characterized in that... Includes the following steps: S1. Synchronously collect real-time load parameters and multi-point temperature parameters from the controller, bind the timestamp to generate parameter pair sequences, and output the load mean and temperature extreme value statistical features through a sliding window buffer. Only the two types of statistical features are used as the sole input for subsequent calorific value matrix and temperature field mapping modeling, discarding the instantaneous original parameters. S2. Map statistical characteristics to the power module loss map, project load along the loss trajectory to generate a three-dimensional heat generation distribution matrix; simultaneously fit the global temperature gradient field, screen out abnormal gradient regions at the heat dissipation channel outlets as heat dissipation bottlenecks, and then... Calculate the characteristic value of the bottleneck thermal resistance, where, The characteristic value of thermal resistance in the heat dissipation bottleneck area, in °C / W; The steady-state temperature difference between the two ends of the heat dissipation bottleneck area, in °C. The steady-state heat flux through the heat dissipation bottleneck region, in W; S3. Establish a three-dimensional mesh coordinate system matching the physical topology of the controller, construct heat-generating and heat-dissipating layer meshes layer by layer, calculate the interlayer thermal coupling coefficient by combining mesh spacing and medium thermal conductivity, and map heat generation intensity and thermal resistance to the corresponding meshes to construct a thermal-energy coupling model. Configure heat generation intensity and heat dissipation capacity fields for each mesh. ,in, The heat dissipation capacity of the grid represents the maximum heat dissipation capacity that the grid can withstand, expressed in W / ℃. The characteristic value of thermal resistance in the heat dissipation bottleneck area, in °C / W; The reference thermal conductivity area corresponding to the grid, in meters. 2 ; S4. Normalize the grid heat intensity and heat dissipation capacity to the global maximum value, and mark the grid with normalized heat intensity greater than heat dissipation capacity as thermal mismatch nodes; search for surrounding redundant heat dissipation grids by Manhattan distance, plan the shortest virtual heat flow conduction path without negative capacity, and adaptively match the corresponding time window according to the path thermal resistance and thermal capacity response time and thermal mismatch over-limit amplitude, and encapsulate the path and window parameters into dynamic thermal constraints. S5. Match the switching sequence power devices and thermal mismatch module, mark the corresponding high-temperature switching events; split the heating pulses equally according to the number of windows and disperse them into each response window to generate a recombined switching sequence; convert it into a drive level sequence with device dead zone interval as a temperature control command; S6. Send temperature control commands to the drive circuit, using the target temperature setpoint of the power device as the control target for non-electrical physical quantity adjustment, periodically collect the actual temperature and compare it with the target temperature setpoint; if the actual temperature is too high, reduce the switching duty cycle to reduce heat generation; if the actual temperature is too low, maintain the switching duty cycle unchanged and only finely adjust the switching frequency; through closed-loop feedback adjustment, the actual temperature converges to the target temperature setpoint.

2. The method according to claim 1, characterized in that, The specific steps for generating the three-dimensional heat distribution matrix in S2 are as follows: Step 1: Extract two types of load values: real-time on-state current and switching frequency of a single power module; Step 2: Project the on-state current along the on-state loss trajectory to the module terminals, solder joints, and internal cable mesh, and accumulate them to obtain the on-state heating base value; Step 3: Project the switching frequency along the switching loss trajectory onto the module driver gate and parasitic inductance grid, and sum them to obtain the base value of switching heating. Step 4: Overlay the on-state and switching heating base values ​​within the same 3D mesh, and fill the corresponding coordinates of the heat distribution matrix with the overlay result as the mesh heating intensity.

3. The method according to claim 1, characterized in that, The specific steps for extracting the heat dissipation bottleneck area in S2 are as follows: Step 1: Read the coordinates of all temperature measurement points and the measured temperature values ​​of the casing and power devices in the temperature field distribution diagram; Step 2: Based on the spatial location and temperature value of each measuring point, calculate the magnitude of the temperature gradient vector grid by grid. Step 3: Define the grid with a gradient modulus greater than the preset gradient threshold of 1.5℃ / cm as the temperature gradient anomaly zone; Step 4: Select the mesh set located at the heat sink airflow outlet and the edge of the thermally conductive substrate within the abnormal area as the heat dissipation bottleneck, record the bottleneck boundary coordinates and substitute them into the equation. Calculate the thermal resistance.

4. The method according to claim 1, characterized in that, The specific steps for constructing the thermal-energy coupling model in S3 are as follows: Step 1: Divide the 3D mesh into X / Y axes to match the length and width of the controller, and divide the Z axis into power device heat dissipation layer and housing heat dissipation layer; Step 2: Fill the heating intensity of all grids in the heat generation distribution matrix into the heating layer coordinate by coordinate; map the heat dissipation bottleneck thermal resistance to the corresponding grid of the heat dissipation layer according to the boundary coordinates; Step 3: Take the straight-line distance between the centers of the vertically aligned heat-generating and heat-dissipating grids, and the preset thermal conductivity of the heat-conducting medium between the grids is 2.5W / (m·K), and calculate the interlayer thermal coupling coefficient; Step 4: Use the thermal coupling coefficient to associate the upper and lower layers of meshes with the same coordinates to complete the binding of the two-layer meshes and form a thermal-energy coupled association model.

5. The method according to claim 1, characterized in that, The specific steps for S4 to identify thermal mismatch nodes are as follows: Step 1: Traverse all 3D meshes of the model and retrieve the original heat intensity and original heat dissipation capacity values ​​for each cell; Step 2: Extract the maximum heat intensity and maximum heat dissipation capacity of the entire mesh as normalization benchmarks; Step 3: Calculate the dimensionless comparison values ​​respectively: Normalized heat generation value = heat generation intensity of a single cell / maximum heat generation intensity of the entire area, Normalized heat dissipation value = heat dissipation capacity of a single cell / maximum heat dissipation capacity of the entire area; Step 4: Compare the two sets of dimensionless values. If the normalized calorific value is larger, mark the current grid as a thermal mismatch node, indicating an exceedance of the limit. ,in, The thermal mismatch node's overheating exceeds the limit, dimensionless, representing the relative proportion of mesh heating exceeding heat dissipation capacity; The original heat intensity of a single grid, in W. This represents the original heat dissipation capacity of a single grid, in W / ℃.

6. The method according to claim 1, characterized in that, The specific steps for planning the virtual heat flow conduction path and response window in S4 are as follows: Step 1: Using the thermal mismatch mesh as the center, search all surrounding meshes within a preset maximum radius of 8 squares from near to far according to the Manhattan distance; Step 2: Select grids that simultaneously meet the conditions of normalized heat dissipation value > normalized heat generation value and grid heat dissipation capacity > 0 as redundant heat dissipation nodes; Step 3: Connect the grid edges of redundant nodes to mismatched nodes to generate multiple candidate path chains, and eliminate invalid paths that pass through grids with heat dissipation capacity ≤ 0; Step 4: Select the candidate path with the fewest grid nodes as the optimal virtual heat flow conduction path; Step 5: Substitute the path equivalent thermal resistance and equivalent heat capacity to solve for the first-order system step response time. Combine the thermal mismatch over-limit amplitude adaptive correction time and set the corrected time as the path-specific response time window length.

7. The method according to claim 1, characterized in that, The specific steps for handling the S5 high-temperature switch disassembly event are as follows: Step 1: Read the original switch sequence one by one, and record the trigger time and the bound power device number of each switch event; Step 2: Match the power device number with the power module to which the thermal mismatch node output by S4 belongs. If the match is successful, mark the event as a high thermal switching event. Step 3: Extract the total heating pulse area and total duration of the high-thermal-switching event, and read the total number N of response windows within the dynamic thermal constraint; Step 4: Divide the original heating pulse area into N equal parts to generate N sub-switch events, with each sub-switch allocated the same pulse area; Step 5: Insert the N sub-switch events into the N independent response time windows at uniform time intervals, and output the recombined switch sequence.

8. The method according to claim 7, characterized in that, The specific steps for S5 to generate the temperature control command are as follows: Step 1: Traverse all sub-switch events in the recombination switch sequence and generate corresponding drive pin level transition signals. The rising edge is the device turn-on time, the falling edge is the turn-off time, and the interval between the rising and falling edges is equal to the sub-switch conduction time. Step 2: Extract the measured turn-on delay and turn-off delay of the power device, and take the larger of the two values ​​as the preset device dead time reference of 0.8μs; Step 3: Insert a 0.8μs dead interval between two adjacent sets of level transition signals; Step 4: Connect all the level transition signals in sequence according to time to form a continuous drive level sequence, which serves as the temperature control command sent to the drive circuit to regulate the temperature rise of the device.

9. The method according to claim 1, characterized in that, The specific operation steps for the S6 closed-loop temperature control convergence regulation are as follows: Step 1: Collect the real-time temperature of the power devices inside the controller every 100ms preset acquisition period, and read the preset target temperature of the temperature rise envelope at the current moment; Step 2: Perform a difference calculation to obtain the temperature deviation. A deviation > 0 indicates that the actual temperature exceeds the standard. Step 3: When the deviation is greater than 0, output a duty cycle reduction command to the drive circuit to synchronously reduce the proportion of single-switch cycle conduction time according to the deviation ratio. Step 4: When the deviation is ≤0, keep the current duty cycle unchanged, and only slightly adjust the switching frequency within the ±1kHz preset frequency adjustment range to improve the dynamic response of the device; Step 5: Repeatedly execute the data acquisition, comparison, and control process to gradually reduce the absolute value of the deviation between the actual temperature and the target temperature until stable convergence.

10. The method according to claim 1, characterized in that, The specific operation steps of the S1 sliding window preprocessing parameters on the sequence are as follows: Step 1: Input all load and temperature parameters with uniform timestamps into a fixed-width sliding window buffer; Step 2: The cache executes the first-in, first-out rule, and when a new parameter pair enters, it automatically removes the set of parameters that has been stored in the window the longest. Step 3: After the window is filled, calculate the arithmetic mean of all load parameters and the maximum value of all temperature parameters within the window in real time. Step 4: Output the average load and maximum temperature as the only input data to the S2 loss map and temperature field mapping process.

Images