Intelligent temperature control system and method of linear servo actuator

Abstract

The embodiment of the invention provides an intelligent temperature control system and method for a linear servo actuator, and the method comprises the steps: obtaining the surface temperature distribution data of a power device through embedding a first temperature sensing array in a circuit board of the linear servo actuator, and capturing the temperature gradient data of a heat dissipation channel through arranging a second temperature sensing array on the inner wall of a housing; inputting the data, the real-time load current and the motion frequency into a heat conduction model to generate a future temperature trend prediction result, selecting an active heat dissipation mode of a corresponding level from a preset heat dissipation strategy library according to the highest temperature point position and the temperature rise rate in the future temperature trend prediction result, sequentially activating a heat dissipation assembly set according to the active heat dissipation mode, and performing heat dissipation according to the active heat dissipation mode. And temperature feedback data are continuously acquired to adjust the heat dissipation intensity, so that intelligent and efficient temperature control of the linear servo actuator is realized, and the stability and reliability of equipment are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial automation control, in particular to an intelligent temperature control system and method for a linear servo actuator. BACKGROUND

[0002] In the field of modern industrial automation control, as a key motion control component, the running stability and reliability of the linear servo actuator directly affect the performance of the entire system. With the development of industrial equipment towards high precision, high speed and high load, a large amount of heat will be generated by the power devices inside the linear servo actuator during operation.

[0003] At present, for the temperature control of the linear servo actuator, the common method is to use a single temperature sensor to monitor the temperature of the key position, and to perform simple heat dissipation control based on the temperature value, such as turning on or off the fan. However, this single temperature monitoring method has obvious limitations. Due to the complex internal structure of the linear servo actuator and the uneven distribution of power devices, there may be a large difference in temperature at different positions, and a single temperature sensor cannot comprehensively and accurately reflect the temperature distribution of the entire actuator. In addition, relying only on the current temperature for heat dissipation control cannot predict the temperature trend in advance, which may lead to untimely or excessive heat dissipation, affecting the performance and life of the actuator, and even may cause damage to the power devices due to high temperature, leading to system failure. SUMMARY

[0004] Therefore, the purpose of the embodiments of the present application is to provide an intelligent temperature control system and method for a linear servo actuator.

[0005] According to one aspect of the embodiments of the present application, an intelligent temperature control method for a linear servo actuator is provided, which comprises: embedding a first temperature sensor array on the circuit board of the linear servo actuator to obtain surface temperature distribution data of the power devices, and setting a second temperature sensor array on the inner wall of the shell of the linear servo actuator to capture temperature gradient data of the heat dissipation channel; inputting the surface temperature distribution data and the temperature gradient data into a heat conduction model, combining the real-time load current and the motion frequency to generate a future temperature trend prediction result; selecting a corresponding level of active heat dissipation mode from a pre-set heat dissipation strategy library according to the highest temperature point position and the temperature rise rate in the future temperature trend prediction result; activating a set of heat dissipation components in turn according to the active heat dissipation mode, and continuously collecting temperature feedback data of the power devices during the heat dissipation process to adjust the heat dissipation intensity; When it is detected that the temperature of a specific power device exceeds a material tolerance threshold, a parallel circuit switching operation is triggered to divert the current load of the specific power device to a backup element group.

[0006] According to another aspect of the embodiments of the present application, there is provided an intelligent temperature control system of a linear servo actuator, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete communication with each other through the communication bus; the memory is used to store a computer program; and the processor is used to execute the computer program to implement the steps of the intelligent temperature control method of the linear servo actuator according to any one of the above aspects.

[0007] According to another aspect of the embodiments of the present application, there is provided a readable storage medium, which stores a computer program, and the computer program can execute the steps of the intelligent temperature control method of the linear servo actuator when run by a processor.

[0008] According to any one of the above aspects, the embodiments of the present application embed a first temperature sensing array on a circuit board of a linear servo actuator to obtain surface temperature distribution data of power devices, and set a second temperature sensing array on an inner wall of a shell to capture temperature gradient data of a heat dissipation channel, combine the surface temperature distribution data and the temperature gradient data with real-time load current and motion frequency to input a heat conduction model to generate a future temperature trend prediction result, realize early prediction of temperature change, and avoid the hysteresis of the traditional method which only controls according to the current temperature. According to the future temperature trend prediction result, a corresponding level of active heat dissipation mode is selected from a preset heat dissipation strategy library, and the heat dissipation component set is activated in sequence according to the active heat dissipation mode, while temperature feedback data is continuously collected to adjust the heat dissipation intensity. This dynamic and accurate heat dissipation control method can effectively improve the heat dissipation efficiency, reduce energy consumption, and ensure that the linear servo actuator can stably operate under various working conditions. In addition, when it is detected that the temperature of a specific power device exceeds a material tolerance threshold, a parallel circuit switching operation is triggered to divert the current load to a backup element group, further enhancing the reliability and safety of the actuator, prolonging the service life of the equipment, and reducing the maintenance cost.

[0009] In order to make the above-mentioned purposes, features and advantages of the embodiments of the present application more obvious and easy to understand, the following will be described in detail with reference to the embodiments and in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some of the embodiments of the present application, and therefore should not be regarded as a limitation to the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0011] Figure 1 The assembly schematic diagram of the intelligent temperature control system of the linear servo actuator provided by the embodiment of the present application is shown. Figure 2 The flow schematic diagram of the intelligent temperature control method of the linear servo actuator provided by the embodiment of the present application is shown. DETAILED DESCRIPTION

[0012] In order to make the person skilled in the art better understand the present application, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. According to the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0013] The terms "first", "second", "third" and the like (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0014] Figure 1An example component diagram of the intelligent temperature control system 100 of a linear servo actuator is shown. The intelligent temperature control system 100 of a linear servo actuator can include one or more processors 104, such as one or more central processing units (CPUs), each of which can implement one or more hardware threads. The intelligent temperature control system 100 of a linear servo actuator can also include any storage media 106 for storing any kind of information, such as code, settings, data, etc. Without limitation, for example, the storage media 106 can include any one or combination of the following: any type of RAM, any type of ROM, a flash memory device, a hard disk, an optical disk, etc. More generally, any storage media can store information using any technology. Further, any storage media can provide volatile or non-volatile retention of information. Further, any storage media can represent a fixed or removable component of the intelligent temperature control system 100 of a linear servo actuator. In one case, the intelligent temperature control system 100 of a linear servo actuator can perform any of the operations described in association with the dependent instructions stored in any storage media or combination of storage media when the processors 104 execute the instructions. The intelligent temperature control system 100 of a linear servo actuator also includes one or more drive units 108, such as a hard disk drive unit, an optical disk drive unit, etc., for interacting with any storage media.

[0015] The intelligent temperature control system 100 of a linear servo actuator also includes an input / output 110 (I / O) for receiving various input (via input unit 112) and for providing various output (via output unit 114). One particular output mechanism can include a presentation device 116 and a presence-dependent graphical user interface (GUI) 118. The intelligent temperature control system 100 of a linear servo actuator can also include one or more network interfaces 120 for exchanging data with other devices via one or more communication units 122. One or more communication buses 124 couple the above-described components together.

[0016] The communication units 122 can be implemented in any way, for example, through a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication units 122 can include any combination of hardwired links, wireless links, routers, gateway functionality, name servers, etc., governed by any protocol or combination of protocols.

[0017] Figure 2 A flow diagram of the intelligent temperature control system and method of a linear servo actuator provided by embodiments of the present invention is shown, which can be executed by the intelligent temperature control system 100 of a linear servo actuator shown in Figure 1 The detailed steps of the intelligent temperature control method of a linear servo actuator are introduced as follows.

[0018] Step S110: embedding a first temperature sensor array on the circuit board of the linear servo actuator to obtain surface temperature distribution data of the power device, and setting a second temperature sensor array on the inner wall of the shell of the linear servo actuator to capture temperature gradient data of the heat dissipation channel.

[0019] In this embodiment, the linear servo actuator's power devices generate heat during operation. Accurately understanding the surface temperature distribution of the power devices and the temperature gradient in the heat dissipation channels is crucial for intelligent temperature control. The first temperature sensor array must be appropriately embedded based on the layout and heat generation characteristics of the power devices on the circuit board. Power devices exhibit varying heat generation patterns, with some areas experiencing concentrated heat and others experiencing more dispersed heat. Therefore, it is crucial to ensure that the first temperature sensor array comprehensively and accurately covers the surface of each power device to obtain accurate surface temperature distribution data.

[0020] Step S111: the obtaining of surface temperature distribution data of the power device includes arranging a micro thermocouple array on the surface of each power device of the circuit board, wherein the detection point density of the thermocouple array is positively correlated with the heating intensity of the device.

[0021] When placing micro-thermocouple arrays on the surfaces of various power devices on a circuit board, it's important to consider the differences in heat generation intensity among these devices. Power devices with high heat generation intensity generate more concentrated heat, necessitating a higher density of detection points for more accurate temperature measurement. For example, if power device A generates more heat than power device B, the density of detection points in the micro-thermocouple array placed on the surface of power device A should be greater than that of power device B. This allows for more detailed capture of subtle changes in the surface temperature of the power devices.

[0022] Step S112: synchronously collecting instantaneous temperature values ​​of multiple detection points through the thermocouple array, and calculating the temperature change slope between adjacent detection points.

[0023] After the micro-thermocouple array is arranged, the instantaneous temperature values ​​of multiple detection points are collected synchronously. These instantaneous temperature values ​​reflect the temperature conditions of each detection point at a specific moment. Next, the temperature change slope between adjacent detection points is calculated. Suppose the adjacent detection points are P1 and P2, and their corresponding instantaneous temperature values ​​are T1 and T2 respectively, and the distance between the two points is d, then the temperature change slope k=(T2-T1) / d. The temperature change slope can reflect the rate of change of temperature between adjacent detection points, which helps to find areas where the temperature changes sharply.

[0024] Step S113: performing spatial interpolation processing on the instantaneous temperature value and the temperature change slope to generate a two-dimensional temperature field map reflecting the hot spot distribution on the surface of the power device.

[0025] After obtaining the instantaneous temperature value and the temperature change slope, spatial interpolation processing is performed. Spatial interpolation is a method of estimating the data of unknown points based on the data of known points. By analyzing and processing the instantaneous temperature value and the temperature change slope of each detection point, the temperature value of other positions on the surface of the power device can be estimated using appropriate interpolation algorithms (such as bilinear interpolation, Kriging interpolation, etc.). Integrating these estimated temperature values, a two-dimensional temperature field map reflecting the hot spot distribution on the surface of the power device can be generated. The two-dimensional temperature field map can intuitively show the distribution of the temperature on the surface of the power device, and the hot spot area will be represented by a higher temperature value in the map.

[0026] Step S114: identifying a local temperature peak area based on the two-dimensional temperature field map, and marking the device number and position coordinates corresponding to the local temperature peak area.

[0027] According to the generated two-dimensional temperature field map, a local temperature peak area is identified. The local temperature peak area refers to an area with a temperature significantly higher than the surrounding areas in the map. A temperature threshold can be set, and when the temperature of a certain area exceeds the threshold, it is determined as a local temperature peak area. Then, the device number and position coordinates corresponding to the local temperature peak area are marked. The device number is used to clearly identify which power device has a local high temperature, and the position coordinates can accurately locate the specific position of the high temperature area on the surface of the power device, facilitating subsequent heat dissipation processing.

[0028] For the second temperature sensing array, it needs to be reasonably arranged along the direction of the heat dissipation channel when installed on the inner wall of the shell. The heat dissipation channel is the channel through which heat is transferred from the power device to the outside, and the temperature at different positions may differ. The second temperature sensing array needs to be able to monitor these temperature differences in real time to obtain temperature gradient data of the heat dissipation channel. By analyzing the temperature gradient data, the direction and speed of heat transfer in the heat dissipation channel can be understood.

[0029] Step S120: inputting the surface temperature distribution data and the temperature gradient data into a heat conduction model, combining real-time load current and motion frequency to generate a future temperature trend prediction result.

[0030] In this embodiment, the surface temperature distribution data of the power device and the temperature gradient data of the heat dissipation channel have been obtained, and next these data will be input into a heat conduction model, while combining real-time load current and motion frequency to predict future temperature trend. The heat conduction model is a mathematical model established based on the principle of heat conduction, which can describe the process of heat transfer within an object.

[0031] Step S121: Establish a finite element analysis model containing heat conduction equations and convection boundary conditions, and map the surface temperature distribution data and the temperature gradient data into initial temperature field input.

[0032] First, a finite element analysis model is established, which contains heat conduction equations and convection boundary conditions. The heat conduction equation describes the conduction law of heat within the object, and the convection boundary condition considers the heat exchange between the object and the surrounding environment. The previously obtained surface temperature distribution data and temperature gradient data are processed so that they can be mapped into the initial temperature field input of the finite element analysis model. Specifically, the surface temperature distribution data and the temperature gradient data are distributed according to the grid division method of the model, so that each grid node has a corresponding initial temperature value. In this way, the finite element analysis model can simulate the heat transfer process based on these initial conditions.

[0033] Step S122: Convert the real-time load current into a heat source power density function, and convert the motion frequency into a heat dissipation boundary condition update frequency parameter.

[0034] The real-time load current is an important parameter when the linear servo actuator is running, which is closely related to the heat generation of the power device. The real-time load current is converted into a heat source power density function, which describes the rate of heat generation per unit volume of the power device. For example, assuming that the real-time load current is I, it can be converted into a heat source power density q through certain physical relationships (such as Joule's law). The motion frequency reflects the working state of the linear servo actuator, which will affect the airflow in the heat dissipation channel and thus affect the heat dissipation effect. The motion frequency is converted into a heat dissipation boundary condition update frequency parameter, that is, the update frequency of the heat dissipation boundary condition is adjusted according to the change of the motion frequency. When the motion frequency is high, the update frequency of the heat dissipation boundary condition is also increased accordingly to more accurately simulate the heat dissipation process.

[0035] Step S123: Superimpose a long short-term memory network prediction module trained by historical temperature data in the finite element analysis model, and the long short-term memory network prediction module outputs the temperature change curve of each power device in the future time window.

[0036] To improve the accuracy of temperature prediction, a long short-term memory network prediction module is superimposed in the finite element analysis model. The long short-term memory network (LSTM) is a neural network capable of processing sequence data, which can learn the long-term dependencies in the data. The LSTM is trained using historical temperature data to capture the regularity of temperature changes. The trained LSTM prediction module can output the temperature change curve of each power device in the future time window according to the current input data (including surface temperature distribution data, temperature gradient data, real-time load current and motion frequency, etc.). The temperature change curve reflects the temperature change trend of each power device in the future period of time.

[0037] Step S124: Extract the maximum temperature rise rate and hotspot migration path according to the temperature change curve, and generate the future temperature trend prediction result containing time markers.

[0038] The maximum temperature rise rate and hotspot migration path are extracted from the temperature change curve output by the long short-term memory network prediction module. The maximum temperature rise rate represents the fastest rate of temperature rise of the power device in the future time window, which reflects the intensity of heat generation of the power device. The hotspot migration path describes the moving track of the temperature peak area on the surface of the power device. By analyzing the temperature change curve, the time period with the fastest temperature rise and the corresponding temperature rise rate, as well as the moving direction and distance of the temperature peak area, are found. The maximum temperature rise rate, hotspot migration path and corresponding time information are integrated to generate the future temperature trend prediction result containing time markers. The future temperature trend prediction result can clearly show the temperature change of each power device in the future period of time.

[0039] Step S130: According to the highest temperature point position and temperature rise rate in the future temperature trend prediction result, select the corresponding level of active cooling mode from the preset cooling strategy library.

[0040] In this embodiment, the future temperature trend prediction result is obtained, which contains important information such as the position of the highest temperature point and the temperature rise rate. According to these information, the appropriate active cooling mode is selected from the preset cooling strategy library to ensure that the temperature of the linear servo actuator is within the safe range.

[0041] Step S131: Construct a three-level cooling response mechanism, wherein the first level response is associated with natural convection cooling condition, the second level response is associated with forced air cooling condition, and the third level response is associated with semiconductor refrigeration and power load reduction condition.

[0042] The three-level heat dissipation response mechanism is constructed to take different intensity of heat dissipation measures according to different temperature conditions. The first level response corresponds to the natural convection heat dissipation condition. When the temperature is low, natural convection can meet the heat dissipation requirement. Natural convection refers to the natural flow of fluid caused by temperature difference, and heat is transferred to the surrounding environment through this natural flow. The second level response is associated with forced air cooling heat dissipation condition. When the temperature rises to a certain extent, only natural convection cannot meet the heat dissipation requirement, at this time the forced air cooling system is started, and the air flow is accelerated through the fan and other equipment to improve the heat dissipation efficiency. The third level response is associated with semiconductor refrigeration and power load reduction condition. When the temperature is too high, not only the semiconductor refrigeration device is started for refrigeration, but also the load of power device is reduced to reduce the heat generation. When the servo actuator adopts the built-in design scheme, when the temperature rises to a certain extent, only natural convection cannot meet the heat dissipation requirement, at this time the internal temperature control system is started, and the internal device efficiency is reduced to reduce the heat output and improve the heat dissipation efficiency. The third level response is associated with semiconductor refrigeration and power load reduction condition. When the temperature is too high, the load of power device needs to be automatically switched off to reduce the heat generation.

[0043] Step S132: comparing the highest temperature point in the future temperature trend prediction result with the preset temperature threshold sequence, activating the first level response when the highest temperature point reaches the first threshold, activating the second level response when the highest temperature point reaches the second threshold, and activating the third level response when the highest temperature point reaches the third threshold.

[0044] The preset temperature threshold sequence includes three thresholds, corresponding to three levels of heat dissipation response mechanisms. The highest temperature point in the future temperature trend prediction result is compared with these thresholds. When the highest temperature point reaches the first threshold, it means that the temperature starts to rise, but is still at a relatively low level, at which time the first level of response is activated to rely on natural convection for heat dissipation. When the highest temperature point reaches the second threshold, it indicates that the temperature is further rising, and natural convection alone cannot meet the heat dissipation demand, at which time the second level of response is activated to start the forced air cooling system. When the highest temperature point reaches the third threshold, it means that the temperature is already very high, and more powerful heat dissipation measures need to be taken, and the third level of response is activated to start the semiconductor refrigeration device and reduce the load of the power device. When the servo actuator adopts a built-in design scheme, when the highest temperature point reaches the first threshold, it means that the temperature starts to rise, but is still at a relatively low level, at which time the first level of response is activated to rely on natural convection for heat dissipation. When the highest temperature point reaches the second threshold, it indicates that the temperature is further rising, and natural convection alone cannot meet the heat dissipation demand, at which time the internal temperature control system is started to reduce the heat output by reducing the internal device efficiency to improve the heat dissipation efficiency. When the highest temperature point reaches the third threshold, it means that the temperature is already very high, and more powerful heat dissipation measures need to be taken, and the third level of response is activated, which is associated with the semiconductor refrigeration and power load reduction conditions. When the temperature is too high, the power device load needs to be automatically switched off to reduce heat generation.

[0045] Step S133: Adjust the working direction of the heat dissipation assembly according to the hot spot migration path, so that the heat dissipation airflow or the refrigeration area covers the expected position of the hot spot migration path.

[0046] The hot spot migration path reflects the moving track of the temperature peak area on the surface of the power device. According to the hot spot migration path, the working direction of the heat dissipation assembly is adjusted. For example, if the hot spot migration path shows that the temperature peak area will move in a certain direction, the direction of the fan or the refrigeration area of the semiconductor refrigeration device is adjusted so that it can cover the expected position of the hot spot migration path. In this way, heat dissipation can be more targeted, and heat dissipation efficiency can be improved.

[0047] Step S140: Activate the set of heat dissipation assemblies in sequence according to the active heat dissipation mode, and continuously collect temperature feedback data of the power device during the heat dissipation process to adjust the heat dissipation intensity.

[0048] In this embodiment, a suitable active heat dissipation mode has been selected, and then the set of heat dissipation assemblies is activated in sequence according to the active heat dissipation mode, and the temperature feedback data of the power device is collected in real time during the heat dissipation process, and the heat dissipation intensity is adjusted according to the feedback data to achieve precise temperature control.

[0049] Step S141: Monitor the current rotating speed of the cooling fan and the airflow pressure data under the condition of forced air cooling, and calculate the actual coverage rate of the airflow in the cooling channel.

[0050] Under the condition of forced air cooling, the cooling fan is the main cooling component. Monitor the current rotating speed of the cooling fan and the airflow pressure data, which reflect the working state of the fan. Through the analysis of the airflow pressure data, the distribution of the airflow in the cooling channel can be understood. Calculate the actual coverage rate of the airflow in the cooling channel, that is, the proportion of the area covered by the airflow to the total area of the cooling channel. Let the total area of the cooling channel be S, and the area covered by the airflow be S1, then the actual coverage rate=S1 / S. The actual coverage rate can reflect the effect of air cooling.

[0051] Step S142: Generate a first adjustment signal according to the difference value between the actual coverage rate and the expected coverage rate, the first adjustment signal being used to control the rotating angle and rotating speed combination of the fan array.

[0052] Compare the calculated actual coverage rate with the expected coverage rate to obtain a difference value. The expected coverage rate is a target value set in advance according to the cooling demand. Generate a first adjustment signal according to the difference value, which is used to control the rotating angle and rotating speed combination of the fan array. If the actual coverage rate is less than the expected coverage rate, it means that the air cooling effect is not good, and the rotating angle and rotating speed of the fan need to be adjusted to make the airflow better cover the cooling channel. For example, increase the rotating speed of the fan or adjust the rotating angle of the fan to improve the coverage rate of the airflow.

[0053] Step S143: Monitor the working current of the refrigeration sheet and the temperature gradient of the refrigeration surface under the condition of semiconductor refrigeration, and generate a second adjustment signal according to the difference value between the temperature gradient and the target cooling rate.

[0054] Under the condition of semiconductor refrigeration, the refrigeration sheet is the key cooling component. Monitor the working current of the refrigeration sheet and the temperature gradient of the refrigeration surface, the working current affects the refrigeration capacity of the refrigeration sheet, and the temperature gradient reflects the refrigeration effect. Compare the monitored temperature gradient with the target cooling rate to obtain a difference value. The target cooling rate is a desired cooling speed set according to the cooling demand. Generate a second adjustment signal according to the difference value, which is used to adjust the working state of the refrigeration sheet, such as increasing or decreasing the working current, to achieve the target cooling rate.

[0055] Step S144: Input the first adjustment signal and the second adjustment signal into the cooling control loop to dynamically adjust the power distribution ratio of the cooling components to achieve temperature balance.

[0056] The first adjustment signal and the second adjustment signal are input into a heat dissipation control loop. The heat dissipation control loop dynamically adjusts the power distribution ratio of the heat dissipation components according to these signals. For example, when the first adjustment signal shows that the air cooling heat dissipation intensity needs to be increased, the heat dissipation control loop will increase the power of the fan; when the second adjustment signal shows that the refrigeration effect needs to be improved, the power of the refrigeration fin will be increased. By dynamically adjusting the power distribution ratio, the various heat dissipation components work together to achieve temperature balance of the power device.

[0057] Step S150: When it is detected that the temperature of a specific power device exceeds the material tolerance threshold, trigger the parallel circuit switching operation to transfer the current load of the specific power device to the backup element group.

[0058] In this embodiment, the temperature of the power device is continuously monitored during heat dissipation. When it is detected that the temperature of a specific power device exceeds the material tolerance threshold, it means that the power device is in an overheated state and may be damaged. At this time, the parallel circuit switching operation needs to be triggered to transfer the current load of the specific power device to the backup element group to protect the power device.

[0059] Step S151: When the temperature of the target power device is detected to be abnormal, query the pre-set redundant circuit connection topology, which contains a backup element group in parallel with the target device.

[0060] When the temperature of the target power device is detected to be abnormal, i.e. the temperature exceeds the material tolerance threshold, the pre-set redundant circuit connection topology is queried. The redundant circuit connection topology is a pre-designed circuit structure, which contains a backup element group in parallel with the target device. By querying the redundant circuit connection topology, the position and connection mode of the backup element group can be determined, which prepares for the subsequent current transfer.

[0061] Step S152: Close the backup circuit path through the shape memory alloy contact, and the closing action of the shape memory alloy contact is automatically executed by the surface temperature of the target device triggering the phase change mechanism.

[0062] After the connection mode of the backup element group is determined, the backup circuit path is closed through the shape memory alloy contact. Shape memory alloy is a material with special properties that changes shape when the temperature reaches its phase change critical point. The shape memory alloy contact is provided at the electrode connection point of the target power device. When the surface temperature of the target device reaches the phase change critical point of the shape memory alloy, the contact will change shape, thereby closing the backup circuit path. In this way, the current can be transferred from the target power device to the backup element group.

[0063] Step S153: Real-time monitoring of the temperature change data of the standby component group after switching, and generating a power load reduction instruction to reduce the current value flowing through the branch of the standby component group when the temperature of the standby component group is detected to exceed the safety threshold.

[0064] After the current is transferred to the standby component group, the temperature change data of the standby component group is monitored in real time. The standby component group also generates heat when it bears the current load, and if the temperature is too high, it will also affect its normal work. When the temperature of the standby component group is detected to exceed the safety threshold, a power load reduction instruction is generated. The instruction is used to reduce the current value flowing through the branch of the standby component group, thereby reducing the heat generation of the standby component group and ensuring its safe operation.

[0065] Step S154: Recording the temperature response curve in the current switching process and feeding back the temperature response curve to the heat conduction model to correct the subsequent prediction parameters.

[0066] The temperature response curve in the current switching process is recorded, which reflects the temperature change of the power device and the standby component group during the current transfer process. The temperature response curve is fed back to the heat conduction model, and the heat conduction model corrects the subsequent prediction parameters according to the feedback information. For example, adjusting the heat source power density function or the heat dissipation boundary condition and other parameters, so that the heat conduction model can more accurately predict the temperature change.

[0067] Step S1521: The closing action of the shape memory alloy contact includes setting a bimetallic composite structure at the electrode connection point of the target power device, wherein the first layer is a shape memory alloy material and the second layer is a heat-conducting insulating material.

[0068] A bimetallic composite structure is set at the electrode connection point of the target power device, the first layer uses shape memory alloy material, which has the characteristic of shape change when reaching the phase change critical point. The second layer is a heat-conducting insulating material, which functions to transfer the heat generated by the target power device to the shape memory alloy layer, while preventing current conduction between the two layers to ensure the safety of the circuit.

[0069] Step S1522: When the surface temperature of the target power device reaches the phase change critical point of the shape memory alloy, the first layer material deforms to push the contact arm to physically connect with the standby circuit contact.

[0070] When the surface temperature of the target power device rises to the phase change critical point of the shape memory alloy, the shape memory alloy layer will deform. This deformation will push the contact arm to physically connect with the standby circuit contact, thereby closing the standby circuit path and achieving current transfer.

[0071] Step S1523: After the physical connection is established, verify the pass state of the backup circuit through the current sensor, and collect the contact resistance data to evaluate the connection stability.

[0072] After the backup circuit path is closed, the pass state of the backup circuit is verified through the current sensor. The current sensor can detect whether there is current passing through the backup circuit, so as to determine whether the circuit has been successfully turned on. At the same time, the contact resistance data is collected, and the contact resistance reflects the connection quality between the contacts. If the contact resistance is too large, the connection is unstable, which may cause heating and other problems.

[0073] Step S1524: When the contact resistance exceeds the allowed range, trigger an alarm signal and start the auxiliary mechanical locking device to enhance the contact pressure.

[0074] The collected contact resistance data is compared with the allowed range. If the contact resistance exceeds the allowed range, the contact connection is unstable. At this time, an alarm signal is triggered to remind the operator. At the same time, the auxiliary mechanical locking device is started, which can enhance the pressure between the contacts, thereby reducing the contact resistance and improving the stability of the connection.

[0075] The training process of the long short-term memory network prediction module includes: Step S210: Construct a multi-dimensional training sample set from the time series of sample surface temperature distribution data, the time series of sample temperature gradient data, the sample load current time series, and the sample motion frequency time series, wherein the input feature dimension of each training sample is the sum of the number of hot spots of the sample surface temperature distribution data and the number of channels of the sample temperature gradient data, and the output label dimension is the number of temperature prediction points of each power device in the future time window.

[0076] In this embodiment, the training of the long short-term memory network prediction module is to enable it to accurately predict the future temperature trend. First, sample data is collected, including the time series of sample surface temperature distribution data, the time series of sample temperature gradient data, the sample load current time series, and the sample motion frequency time series. The time series of sample surface temperature distribution data records the temperature distribution of the power device surface at different time points, the time series of sample temperature gradient data reflects the change of temperature gradient in the heat dissipation channel with time, the sample load current time series reflects the load current size of the linear servo actuator at different times, and the sample motion frequency time series shows the time variation law of the motion frequency.

[0077] The time series data is combined into a multi-dimensional training sample set. For each training sample, the input feature dimension is the sum of the number of hot spots in the sample surface temperature distribution data and the number of channels in the sample temperature gradient data. The number of hot spots represents the number of areas with higher temperature in the sample surface temperature distribution data, and the number of channels is the number of different channels involved in the sample temperature gradient data. The output label dimension is the number of temperature prediction points for each power device in the future time window, which is a pre-set time range in which the temperature of each power device needs to be predicted, and the number of prediction points is the number of temperature data that needs to be predicted in the time window.

[0078] Step S211: Construct a bidirectional stacked long short-term memory network, wherein the number of input layer neurons is equal to the input feature dimension, the number of output layer neurons is consistent with the output label dimension, and a temperature change rate weighting factor is embedded in the gating unit of each hidden layer to enhance the time series correlation of hot spot areas.

[0079] When constructing a bidirectional stacked long short-term memory network, the structure and parameters of the network need to be determined. The number of input layer neurons is set to be equal to the input feature dimension, so that the input data can be completely processed by the network. The number of output layer neurons is consistent with the output label dimension, so that the network can output the required temperature prediction results.

[0080] A temperature change rate weighting factor is embedded in the gating unit of each hidden layer. The temperature change rate weighting factor is used to enhance the time series correlation of hot spot areas. The temperature change of hot spot areas often has an important influence on the overall temperature trend. By introducing the temperature change rate weighting factor, the network can pay more attention to the temperature change of hot spot areas, thereby improving the prediction accuracy. For example, when calculating the output of the gating unit, the temperature change rate weighting factor is operated with the corresponding input data, so that the information of hot spot areas is more prominent in the network.

[0081] Step S212: Divide the multi-dimensional training sample set into input feature segments and output label segments according to a pre-set time step, wherein the input feature segment contains normalized temperature distribution and load parameters at the current time and the previous N times, and the output label segment contains normalized temperature prediction distribution at the future M times.

[0082] In order to facilitate network training, the multi-dimensional training sample set needs to be segmented according to a preset time step. The preset time step is a time interval determined according to actual requirements and data characteristics. After segmentation, an input feature segment and an output label segment are obtained. The input feature segment contains the normalized temperature distribution and the load parameter of the current time and the previous N times. Normalization is to eliminate the dimensional differences between different features, so that the data is processed on the same scale. The load parameter includes the sample load current and the sample motion frequency. The output label segment contains the normalized temperature prediction distribution of the future M times, that is, the temperature distribution of each power device in the future period of time needs to be predicted by the network.

[0083] Step S213: The data blocks of the input feature segment and the output label segment are intercepted by using a sliding window, the temperature prediction values of each power device are calculated through the forward propagation of the gated recurrent unit, and a loss function is generated based on the spatial temperature gradient difference between the prediction values and the actual values.

[0084] The data blocks of the input feature segment and the output label segment are intercepted by using a sliding window. The sliding window is a way of intercepting data on time series data, which can intercept data blocks of fixed length at different time positions. The forward propagation is performed by the gated recurrent unit, which is the core component of the long short-term memory network and can process sequence data and remember long-term dependencies in the data. In the forward propagation process of the gated recurrent unit, the temperature prediction values of each power device are calculated according to the input data blocks.

[0085] The spatial temperature gradient difference between the prediction values and the actual values is calculated. The spatial temperature gradient difference reflects the difference between the predicted temperature distribution and the actual temperature distribution in space. Based on the difference, a loss function is generated, which is used to measure the error between the network prediction result and the actual result. For example, the mean square error method can be used to calculate the value of the loss function, so that the smaller the value of the loss function, the closer the network prediction result to the actual situation.

[0086] Step S214: Update the network weight parameters according to the gradient direction of the loss function, and calculate the projection component of the temperature gradient difference in the heat dissipation channel direction after each iteration to dynamically adjust the weight update amplitude of different hidden layers.

[0087] The network weight parameters are updated according to the gradient direction of the loss function. The gradient direction represents the direction in which the loss function value decreases fastest, and by updating the weight parameters along this direction, the value of the loss function can be gradually reduced, thereby improving the prediction performance of the network. After each iteration, the projection component of the temperature gradient difference in the heat dissipation channel direction is calculated. The heat dissipation channel direction is an important direction of heat transfer, and the projection component of the temperature gradient difference in this direction can reflect the accuracy of the network in predicting the temperature change in the heat dissipation channel.

[0088] The weight update amplitude of different hidden layers is dynamically adjusted according to the projection component. If the projection component is large, it indicates that the network has a large error in predicting the temperature change of the heat dissipation channel, and at this time the weight update amplitude of the corresponding hidden layer can be appropriately increased to speed up the learning speed of the network; if the projection component is small, it indicates that the prediction effect of the network in this aspect is good, and the weight update amplitude can be appropriately reduced to avoid excessive adjustment.

[0089] Step S215: When the change rate of the projection component in consecutive iterations is lower than the preset convergence threshold, the network weight parameters are locked and coupled with the heat source power density function in the heat conduction model for verification.

[0090] During network training, the change rate of the projection component is continuously monitored. When the change rate of the projection component in consecutive iterations is lower than the preset convergence threshold, it indicates that the training of the network has approached convergence, and at this time the network weight parameters are locked. Locking the weight parameters means that they are no longer updated to maintain the stability of the network.

[0091] The locked network weight parameters are coupled with the heat source power density function in the heat conduction model for verification. The heat source power density function describes the rate at which the power device generates heat, which is an important part of the heat conduction model. Through coupling verification, the compatibility and consistency of the long short-term memory network prediction module and the heat conduction model can be checked to ensure that they can work together and improve the accuracy of temperature prediction.

[0092] Step S216: The locked network weight parameters are fine-tuned online by using the incremental sequence of the real-time collected heat dissipation channel temperature gradient data and load current data, so that the output temperature trend of the output layer and the transient simulation result error of the heat conduction model are minimized.

[0093] After the network training is completed and the weight parameters are locked, online fine-tuning is still needed. The incremental sequence of the real-time collected heat dissipation channel temperature gradient data and load current data is used for fine-tuning the locked network weight parameters. The incremental sequence reflects the changes of these data over a period of time.

[0094] The goal of fine-tuning is to minimize the output temperature trend of the output layer and the transient simulation result error of the heat conduction model. The transient simulation result of the heat conduction model is the result obtained by simulating the temperature change based on the heat conduction principle. By continuously adjusting the network weight parameters, the output result of the long short-term memory network prediction module is closer to the transient simulation result of the heat conduction model, thereby further improving the accuracy of temperature prediction.

[0095] The method further comprises: Step S310: Establish a heat dissipation performance evaluation index system, which includes a temperature uniformity index, a heat dissipation response delay index, and an energy consumption efficiency index.

[0096] In this embodiment, in order to evaluate the effectiveness of the heat dissipation strategy, a heat dissipation performance evaluation index system needs to be established. This system includes three important indexes, namely the temperature uniformity index, the heat dissipation response delay index, and the energy consumption efficiency index. The temperature uniformity index is used to measure the temperature difference between each power device during the heat dissipation process. If the temperature difference between each power device is small, it means that the temperature uniformity is good and the heat dissipation effect is uniform; on the contrary, if the temperature difference is large, it means that the temperature uniformity is poor and there may be a problem of local overheating.

[0097] The heat dissipation response delay index reflects the time delay between temperature rise and the start of the effective work of the heat dissipation system. The shorter the delay time, the faster the heat dissipation system can respond to temperature changes and take timely heat dissipation measures; the longer the delay time, the more likely it is to cause the power device to be in a high temperature state for a period of time, affecting its performance and life.

[0098] The energy consumption efficiency index measures the energy consumed by the heat dissipation system to achieve the same heat dissipation effect. The higher the energy consumption efficiency, the less energy is consumed while achieving the heat dissipation goal, and the better the energy saving effect.

[0099] Step S311: After each heat dissipation action is completed, calculate the performance evaluation score of the current heat dissipation period according to the temperature feedback data.

[0100] After each heat dissipation action is completed, collect temperature feedback data. The temperature feedback data records the temperature changes of each power device during the heat dissipation process. According to these data, the performance evaluation score of the current heat dissipation period is calculated. For the temperature uniformity index, the dispersion degree of temperature can be evaluated by calculating the variance or standard deviation of the temperature of each power device, and then the temperature uniformity score is obtained. For the heat dissipation response delay index, the delay time can be calculated according to the time of temperature rise and the time of the start of the effective work of the heat dissipation system, and the delay time is compared with the preset standard time to obtain the corresponding score. For the energy consumption efficiency index, the ratio between the energy consumed during the heat dissipation process and the heat dissipation effect achieved can be calculated, and the ratio is compared with the preset standard ratio to obtain the energy consumption efficiency score.

[0101] The scores of the three indexes are weighted and summed according to certain weights to obtain the performance evaluation score of the current heat dissipation period. The weights can be adjusted according to actual needs and importance, for example, if more attention is paid to temperature uniformity, the weight of the temperature uniformity index can be appropriately increased.

[0102] Step S312: When the performance evaluation score is continuously lower than the preset reference value, start the heat dissipation strategy optimization process, and readjust the mode switching threshold and component control parameters in the preset heat dissipation strategy library.

[0103] The performance evaluation score obtained each time is compared with the preset reference value. The preset reference value is a reference value set according to the actual application scene and heat dissipation requirements. When the performance evaluation score is continuously lower than the preset reference value, it indicates that the current heat dissipation strategy is not effective, and the heat dissipation strategy optimization process needs to be started.

[0104] In the heat dissipation strategy optimization process, the mode switching threshold and component control parameters in the preset heat dissipation strategy library are readjusted. The mode switching threshold determines which heat dissipation mode is selected under different temperature conditions. By adjusting these thresholds, the heat dissipation system can respond more flexibly to temperature changes. The component control parameters are used to control the working state of the heat dissipation components, such as the speed of the fan, the working current of the cooling fin, etc. Adjusting these parameters can optimize the performance of the heat dissipation components and improve the heat dissipation efficiency.

[0105] Step S313: The optimized heat dissipation strategy is packaged as a firmware update package and distributed to the same type of actuators for batch strategy synchronization through the wireless communication module.

[0106] After completing the optimization of the heat dissipation strategy, the optimized heat dissipation strategy is packaged as a firmware update package. The firmware update package contains information such as new mode switching thresholds and component control parameters. The firmware update package is distributed to the same type of actuators through the wireless communication module. The wireless communication module can realize wireless transmission of data, making it convenient and fast to send the update package to each actuator.

[0107] After receiving the firmware update package, the same type of actuators automatically perform the update operation to realize batch strategy synchronization. This can ensure that all the same type of actuators use the optimized heat dissipation strategy, improving the heat dissipation performance and stability of the entire system.

[0108] Based on the above steps, the embodiment of the present application realizes the early prediction of temperature change by embedding a first temperature sensor array on the circuit board of the linear servo actuator to obtain the surface temperature distribution data of the power device, and setting a second temperature sensor array on the inner wall of the shell to capture the temperature gradient data of the heat dissipation channel, combining the surface temperature distribution data and the temperature gradient data with the real-time load current and the motion frequency to input the heat conduction model to generate the future temperature trend prediction result, avoiding the hysteresis of the traditional method which only controls according to the current temperature. According to the future temperature trend prediction result, the corresponding level of active cooling mode is selected from the pre-set cooling strategy library, and the cooling component set is activated in turn according to the active cooling mode, while the temperature feedback data is continuously collected to adjust the cooling intensity. This kind of dynamic and accurate cooling control mode can effectively improve the cooling efficiency, reduce the energy consumption, and ensure that the linear servo actuator can operate stably under various working conditions. In addition, when it is detected that the temperature of a specific power device exceeds the material tolerance threshold, the parallel circuit switching operation is triggered to transfer the current load to the standby element group, further enhancing the reliability and safety of the actuator, prolonging the service life of the equipment, and reducing the maintenance cost.

[0109] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here.

[0110] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting, and the scope of the present application is defined by the appended claims rather than the above description, and therefore all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present application.

Claims

1. An intelligent temperature control method for a linear servo actuator, characterized in that: The method comprises: Embedding a first temperature sensing array on a circuit board of the linear servo actuator to obtain surface temperature distribution data of a power device, and providing a second temperature sensing array on an inner wall of a housing of the linear servo actuator to capture temperature gradient data of a heat dissipation channel; Inputting the surface temperature distribution data and the temperature gradient data into a heat conduction model, and combining the real-time load current and motion frequency to generate a future temperature trend prediction result; Selecting a corresponding level of active cooling mode from a preset cooling strategy library according to the maximum temperature point position and temperature rise rate in the future temperature trend prediction result; activating the heat dissipation component set in sequence according to the active heat dissipation mode, and continuously collecting temperature feedback data of the power device during the heat dissipation process to adjust the heat dissipation intensity; When it is detected that the temperature of a specific power device exceeds a material tolerance threshold, a parallel circuit switching operation is triggered to transfer the current load of the specific power device to the standby component group.

2. The intelligent temperature control method for a linear servo actuator according to claim 1, characterized in that: The obtaining of surface temperature distribution data of the power device includes: Arranging a micro thermocouple array on the surface of each power device of the circuit board, wherein the detection point density of the thermocouple array is positively correlated with the heating intensity of the device; Synchronously collecting instantaneous temperature values ​​of multiple detection points through the thermocouple array, and calculating the temperature change slope between adjacent detection points; Performing spatial interpolation processing on the instantaneous temperature value and the temperature change slope to generate a two-dimensional temperature field map reflecting the hot spot distribution on the surface of the power device; A local temperature peak region is identified based on the two-dimensional temperature field map, and the device number and position coordinates corresponding to the local temperature peak region are marked.

3. The intelligent temperature control method for a linear servo actuator according to claim 1, characterized in that: The step of inputting the surface temperature distribution data and the temperature gradient data into a heat conduction model and combining the real-time load current and motion frequency to generate a future temperature trend prediction result includes: Establishing a finite element analysis model including a heat conduction equation and convection boundary conditions, and mapping the surface temperature distribution data and the temperature gradient data into an initial temperature field input; Converting the real-time load current into a heat source power density function, and converting the motion frequency into a heat dissipation boundary condition update frequency parameter; A long short-term memory network prediction module trained with historical temperature data is superimposed on the finite element analysis model, wherein the long short-term memory network prediction module outputs a temperature change curve of each power device within a future time window; The maximum temperature rise rate and the hotspot migration path are extracted according to the temperature change curve, and the future temperature trend prediction result including the time mark is generated.

4. The intelligent temperature control method for a linear servo actuator according to claim 3, characterized in that: The selecting of a corresponding level of active heat dissipation mode from a preset heat dissipation strategy library according to the maximum temperature point position and the temperature rise rate in the future temperature trend prediction result includes: Construct a three-level heat dissipation response mechanism, where the first level response is related to natural convection heat dissipation conditions, the second level response is related to forced air cooling conditions, and the third level response is related to semiconductor refrigeration and power load reduction conditions; Comparing the highest temperature point in the future temperature trend prediction result with a preset temperature threshold sequence, activating a first-level response when the highest temperature point reaches a first threshold, activating a second-level response when the highest temperature point reaches a second threshold, and activating a third-level response simultaneously when the highest temperature point reaches a third threshold; The working direction of the heat dissipation component is adjusted according to the hot spot migration path, so that the heat dissipation airflow or the cooling area covers the expected position of the hot spot migration path.

5. The intelligent temperature control method for a linear servo actuator according to claim 4, characterized in that: The step of sequentially activating the heat dissipation component set according to the active heat dissipation mode and continuously collecting temperature feedback data of the power device to adjust the heat dissipation intensity during the heat dissipation process includes: Under forced air cooling conditions, monitoring the current speed and airflow pressure data of the cooling fan, and calculating the actual coverage rate of the airflow in the cooling channel; generating a first adjustment signal according to a difference between the actual coverage and the expected coverage, wherein the first adjustment signal is used to control a combination of a rotation angle and a rotation speed of the fan array; Under semiconductor refrigeration conditions, monitoring the operating current of the refrigeration plate and the temperature gradient of the refrigeration surface, and generating a second adjustment signal according to the difference between the temperature gradient and the target cooling rate; The first adjustment signal and the second adjustment signal are input into a heat dissipation control loop to dynamically adjust the power distribution ratio of the heat dissipation component to achieve temperature balance.

6. The intelligent temperature control method for a linear servo actuator according to claim 1, characterized in that: When it is detected that the temperature of a specific power device exceeds a material tolerance threshold, triggering a parallel circuit switching operation to transfer the current load of the specific power device to the standby component group includes: When a temperature anomaly of a target power device is detected, querying a preset redundant circuit connection topology, the redundant circuit connection topology including a backup component group connected in parallel with the target device; Closing the alternate circuit path via a shape memory alloy contact, wherein the closing action of the shape memory alloy contact is automatically performed by a phase change mechanism triggered by the surface temperature of the target device; Real-time monitoring of temperature change data of the standby component group after switching, and when it is detected that the temperature of the standby component group exceeds a safety threshold, generating a power load reduction instruction to reduce the current value of the branch flowing through the standby component group; The temperature response curve during the current switching process is recorded, and the temperature response curve is fed back to the heat conduction model to correct subsequent prediction parameters.

7. The intelligent temperature control method for a linear servo actuator according to claim 6, characterized in that: The closing action of the shape memory alloy contact includes: Providing a bimetallic composite structure at the electrode connection point of the target power device, wherein the first layer is a shape memory alloy material and the second layer is a thermally conductive insulating material; When the surface temperature of the target power device reaches the phase transition critical point of the shape memory alloy, the first layer of material deforms to push the contact arm to physically connect with the backup circuit contact; After the physical connection is established, verifying the path status of the backup circuit through a current sensor and collecting contact resistance data to evaluate the connection stability; When the contact resistance exceeds the allowable range, an alarm signal is triggered and an auxiliary mechanical locking device is started to increase the contact pressure.

8. The intelligent temperature control method for a linear servo actuator according to claim 3, characterized in that: The training process of the long short-term memory network prediction module includes: The time series of sample surface temperature distribution data, sample temperature gradient data, sample load current time series, and sample motion frequency time series are used to form a multidimensional training sample set. The input feature dimension of each training sample is the sum of the number of hotspots in the sample surface temperature distribution data and the number of channels in the sample temperature gradient data. The output label dimension is the number of temperature prediction points for each power device in the future time window. Constructing a bidirectional stacked long short-term memory network, wherein the number of neurons in the input layer is equal to the input feature dimension, the number of neurons in the output layer is consistent with the output label dimension, and a temperature change rate weighting factor is embedded in the gating unit of each hidden layer to enhance the temporal correlation of the hot spot area; Segmenting the multidimensional training sample set into input feature segments and output label segments according to a preset time step, wherein the input feature segments include normalized temperature distribution and load parameters at the current moment and the previous N moments, and the output label segments include normalized temperature prediction distribution at the next M moments; A sliding window is used to intercept data blocks of the input feature segment and the output label segment, a temperature prediction value of each power device is calculated through forward propagation of a gated recurrent unit, and a loss function is generated based on the difference in spatial temperature gradient between the predicted value and the actual value; Update the network weight parameters according to the gradient direction of the loss function, and calculate the projection component of the temperature gradient difference in the direction of the heat dissipation channel after each iteration to dynamically adjust the weight update amplitude of different hidden layers; When the rate of change of the projection component in consecutive iterations is lower than a preset convergence threshold, locking the network weight parameters and performing coupling verification on the heat source power density function in the heat conduction model; The locked network weight parameters are fine-tuned online through the incremental sequence of the heat dissipation channel temperature gradient data and load current data collected in real time, so as to minimize the error between the output temperature trend of the output layer and the transient simulation result of the heat conduction model.

9. The intelligent temperature control method for a linear servo actuator according to claim 1, characterized in that: The method further comprises: Establishing a heat dissipation performance evaluation index system, wherein the heat dissipation performance evaluation index system includes a temperature balance index, a heat dissipation response delay index, and an energy efficiency index; After each heat dissipation action is completed, calculating the performance evaluation score of the current heat dissipation cycle based on the temperature feedback data; When the performance evaluation score is continuously lower than a preset benchmark value, a heat dissipation strategy optimization process is initiated to readjust the mode switching threshold and component control parameters in the preset heat dissipation strategy library; The optimized heat dissipation strategy is encapsulated as a firmware update package and distributed to actuators of the same type through the wireless communication module to achieve batch strategy synchronization.

10. An intelligent temperature control system for a linear servo actuator, characterized in that: include: A processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; The memory is used to store a computer program; the processor is used to implement the steps of the intelligent temperature control method for the linear servo actuator according to any one of claims 1 to 9 when executing the computer program.

Images