Multi-source heat coupled solid-state hydrogen storage heat control system, method and medium

Abstract

This invention discloses a multi-source thermally coupled solid-state hydrogen storage thermal control system, method, and medium, comprising: collecting thermally coupled state data and constructing a multi-source thermal state data set; constructing a thermal potential mirror field to generate thermal potential migration trajectories and heat accumulation regions; constructing a reaction relay field to generate reaction relay chains and reaction imbalance regions; constructing a thermal path association structure to identify heat flow succession relationships, heat flow competition relationships, and heat flow blocking relationships, forming a heat flow relay structure; executing multi-source thermal coupling regulation to coordinately adjust heat source output, heat exchange intensity, and cooling medium flow rate; collecting post-regulation state data and updating the multi-source thermal state data set to form a closed-loop thermal control process. This invention achieves directional regulation of solid-state hydrogen storage heat and balanced advancement of hydrogen absorption and desorption reactions through thermal potential prediction, reaction relay analysis, and heat flow path reorganization.

Description

Technical Field

[0001] This invention relates to the field of solid-state hydrogen storage thermal management technology, and in particular to a multi-source thermally coupled solid-state hydrogen storage thermal control system, method and medium. Background Technology

[0002] Solid-state hydrogen storage technology utilizes metal hydrides, composite hydrogen storage materials, or porous hydrogen storage materials to adsorb, absorb, and release hydrogen, offering advantages such as high hydrogen storage density, good safety, and applicability to stationary energy storage systems. During solid-state hydrogen storage, the hydrogen absorption phase is typically accompanied by heat release, while the hydrogen release phase usually requires heat absorption. Therefore, the heat transfer state within the hydrogen storage bed directly affects the hydrogen absorption / desorption rate, hydrogen storage capacity utilization, and system operational stability.

[0003] Existing solid-state hydrogen storage thermal management methods typically use temperature sensors to collect the local temperature of the hydrogen storage bed and regulate the temperature through heaters, heat exchangers, cooling channels, or circulating media. They mostly use temperature deviation as the control basis and mainly achieve heat replenishment or dissipation by adjusting the heat source power, cooling flow rate, or heat exchange intensity, which can improve the problem of excessively high or low hydrogen storage bed temperature to a certain extent.

[0004] However, most existing technologies only focus on temperature feedback, making it difficult to characterize the spatial distribution and migration trend of the driving force of heat conduction inside the hydrogen storage bed, and unable to accurately identify the heat diffusion and heat potential competition relationships under the combined action of multiple heat sources. Existing heat flow paths usually rely on fixed heat exchange structures, which cannot dynamically reorganize the heat transfer path according to the progress of the hydrogen storage reaction, leading to local overheating, insufficient heating, or reaction interruption, which in turn causes uneven progress of the hydrogen storage reaction, decreased hydrogen absorption and desorption efficiency, and insufficient system thermal stability.

[0005] Therefore, how to provide a multi-source thermally coupled solid-state hydrogen storage thermal control system, method, and medium is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] One objective of this invention is to propose a multi-source thermally coupled solid-state hydrogen storage thermal control system, method, and medium. This invention fully utilizes thermally coupled state data sensing, improved wavelet neural operator model, thermal potential mirror field prediction, reaction relay field construction, and heat flow relay structure reorganization technology. It details the realization process of thermal potential migration identification, reaction imbalance location, and multi-source heat directional control in solid-state hydrogen storage, and has the advantages of precise heat distribution, balanced reaction propulsion, high hydrogen absorption and desorption efficiency, and strong system operation stability.

[0007] A multi-source thermally coupled solid-state hydrogen storage thermal control method according to an embodiment of the present invention includes: Collect thermal coupling state data from a solid-state hydrogen storage system, preprocess the thermal coupling state data, and construct a multi-source thermal state data set; A thermal potential mirror field is constructed based on a multi-source thermal state data set. The thermal potential mirror field is input into an improved wavelet neural operator model to identify the thermal diffusion inheritance relationship, thermal potential competition relationship and thermal potential migration direction, and generate the thermal potential migration trajectory and thermal accumulation area within the future time window. Based on thermal potential migration trajectory and thermal accumulation region, a reaction relay field is constructed to identify the reaction transmission relationship, reaction blocking relationship and reaction competition relationship between different regions of the hydrogen storage bed, and to generate reaction relay chain and reaction imbalance region. Based on the reaction relay chain and reaction imbalance region, a heat path association structure is constructed to identify the heat flow inheritance relationship, heat flow competition relationship and heat flow blocking relationship among multiple heat sources. The heat path association relationship is reorganized according to the transmission order of the reaction relay chain to form a heat flow relay structure. Based on the heat flow relay structure, reaction relay chain and reaction imbalance region, multi-source thermal coupling control is performed to coordinate the heat output, heat exchange intensity and cooling medium flow rate of each heat source, so that heat is directionally transferred to the reaction imbalance region along the heat flow relay structure. Real-time acquisition of temperature distribution data, pressure data, and hydrogen absorption / desorption rate data after regulation; updating of multi-source thermal state data set; and reconstruction of thermal potential mirror field, reaction relay field, and heat flow relay structure to form a closed-loop thermal control process.

[0008] Optionally, the thermal coupling status data includes temperature distribution data, pressure data, hydrogen absorption / desorption rate data, heat source input data, and cooling system operation data.

[0009] Optionally, the preprocessing of the thermally coupled state data to construct a multi-source thermal state data set includes: Collect temperature data, pressure data, hydrogen absorption and desorption rate data, heat source input data, and cooling system operation data at different locations in the hydrogen storage bed, and generate corresponding time markers according to the collection time; Time alignment processing is performed on thermally coupled state data with different sampling periods, and a correspondence is established based on a unified time identifier to generate a synchronous thermal state data sequence. Based on the spatial location of each sensor, a spatial coordinate mapping relationship for the hydrogen storage bed is established, and each data item in the synchronous thermal state data sequence is associated with the corresponding spatial location to generate a spatial thermal state data sequence. Abnormal, missing, and duplicate data in the spatial thermal state data sequence are corrected and combined according to time identifier and spatial location to construct a multi-source thermal state data set.

[0010] Optionally, generating the thermal potential migration trajectory and thermal accumulation region within the future time window includes: An improved wavelet neural operator model is constructed, which includes an input lifting layer, a two-branch wavelet transform layer, a cross-branch attention layer, an operator update layer, and an output mapping layer. The input lifting layer and the two-branch wavelet transform layer are connected by a residual connection, the two-branch wavelet transform layer and the cross-branch attention layer are connected by a parallel connection, the cross-branch attention layer and the operator update layer are connected by a serial connection, and the operator update layer and the output mapping layer are connected by a skip connection. Read the thermal coupling state data, write the temperature, heat source power and cooling intensity of different spatial locations of the hydrogen storage bed into the three-dimensional thermal state tensor, and generate the initial thermal potential field with the spatial index as the row and column dimension and the time index as the channel dimension. Based on the initial thermal potential field, the thermal potential gradient direction at each spatial location is calculated. The thermal potential field at the current moment is mapped and associated with the thermal potential fields at each moment within the preset time window to generate a thermal potential mirror field and record the mirror time identifier. The thermal potential mirror field is input into the lifting layer, which uses a one-dimensional convolution operation with a kernel size of three and a channel number of one hundred and twenty-eight to map the thermal potential mirror field to a high-dimensional feature space. High-dimensional features are fed into a two-branch wavelet transform layer, where the local branch performs discrete wavelet transform to extract local multi-scale heat diffusion features, and the global branch performs continuous wavelet transform to extract global thermal potential change features. The outputs of the two branches are fused through residual connection and fed into a cross-branch attention layer. A reaction-coupled attention module is set in the cross-branch attention layer, using the hydrogen absorption and desorption rate and pressure change as external gate control signals to dynamically adjust the weight ratio of local features and global features, and generate a set of thermal potential evolution features. The set of thermal potential evolution features is then sent to the operator update layer, which adopts a three-layer fully connected structure and inserts an adaptive gate control unit to output thermal potential inheritance features, thermal potential competition features, and thermal potential migration direction features. The thermal potential reception characteristics, thermal potential competition characteristics, and thermal potential migration direction characteristics are input into the output mapping layer. The output mapping layer uses a deconvolution structure to generate the thermal potential field at each moment within the future time window. The thermal potential migration trajectory is generated by connecting them in chronological order. The position in the thermal potential migration trajectory where the accumulated thermal potential exceeds the preset aggregation threshold is determined as the thermal aggregation region.

[0011] Optionally, the generation of the reaction relay chain and the reaction imbalance region includes: Read the thermal potential migration trajectory, thermal accumulation region, pressure data and hydrogen absorption and desorption rate data, divide the hydrogen storage bed into multiple reaction units according to the spatial location of the hydrogen storage bed, and write the thermal potential arrival time, thermal potential residence state, thermal potential source region, pressure response state and hydrogen absorption and desorption rate response state for each reaction unit. Based on the thermal potential arrival time, thermal potential residence state, pressure response state and hydrogen absorption / desorption rate response state of each reaction unit, the reaction advancement state of the corresponding reaction unit is determined. The reaction unit that satisfies the continuous arrival of thermal potential, synchronous pressure response and effective change of hydrogen absorption / desorption rate is marked as the reaction receiving unit. The reaction unit that lags in pressure response or has insufficient change in hydrogen absorption / desorption rate after the arrival of thermal potential is marked as the reaction restricted unit. According to the migration direction of the thermal potential migration trajectory, the reaction advancement status between adjacent reaction units is compared sequentially. The adjacent relationship where the previous reaction unit has completed the reaction advancement and the next reaction unit can continue to take over the reaction advancement is marked as a reaction transfer relationship. The adjacent relationship where the next reaction unit cannot take over the reaction advancement of the previous reaction unit is marked as a reaction blocking relationship. The adjacent relationship where multiple reaction units compete for heat in the same thermal accumulation area in the same time slice is marked as a reaction competition relationship. By combining reaction units, reaction propagation states, reaction transmission relationships, reaction blocking relationships, and reaction competition relationships according to time slices and spatial locations, a reaction relay field is constructed, and reaction units with reaction transmission relationships are sequentially connected along the direction of thermal potential migration to generate a reaction relay chain. Read the reaction blocking relationship and reaction competition relationship in the reaction relay field, determine the reaction unit corresponding to the interruption position of the reaction relay chain as the reaction discontinuity unit, determine the reaction unit in the thermal accumulation area whose reaction propagation state has not reached the set reaction propagation conditions as the thermal reaction mismatch unit, merge the spatial range of the reaction discontinuity unit and the thermal reaction mismatch unit to generate the reaction imbalance region.

[0012] Optionally, the reorganization of the thermal path relationships according to the transmission sequence of the reaction relay chain to form a heat flow relay structure includes: Read the reaction relay chain, reaction imbalance region, thermal potential migration trajectory, heat accumulation region and heat source layout data, establish thermal path units according to the hydrogen storage bed spatial partition, and write the heat source source, starting region, target region, heat exchange channel, heat transfer medium, cooling medium, valve status, heat transfer direction and controllable execution object for each thermal path unit; According to the transmission sequence of the reaction relay chain, the positional relationship and heat transfer direction between each heat path unit and the reaction unit are compared. The heat path unit that can connect the previous reaction unit and the next reaction unit is marked as the heat flow receiving unit. The heat path unit that has multiple heat sources acting on the same reaction imbalance region is marked as the heat flow competing unit. The heat path unit whose heat transfer direction is inconsistent with the transmission sequence of the reaction relay chain or cannot reach the next reaction unit is marked as the heat flow blocking unit. The heat flow receiving unit, heat flow competing unit, and heat flow blocking unit are associated according to spatial adjacency, temporal sequence, and heat source relationship to construct a heat path association structure. The receiving target, competing object, blocking position, and adjustable control object corresponding to each heat path unit are recorded in the heat path association structure. Based on the location of the reaction imbalance region and the transmission order of the reaction relay chain, the heat path association structure is reorganized. The heat flow receiving unit pointing to the reaction imbalance region is retained, the output order of the heat source corresponding to the heat flow competition unit is adjusted, the heat flow blocking unit that causes heat deviation is closed or weakened, and the alternative heat path unit that can reach the next reaction unit is matched for the heat flow blocking position. The reorganized thermal path units are connected in the order of heat source initiation area, heat flow receiving area, reaction imbalance area and heat recovery area to form a heat flow relay structure, and output the heat source output control object, heat exchange channel control object, valve opening control object and cooling medium flow control object corresponding to the heat flow relay structure.

[0013] Optionally, the multi-source thermal coupling control based on the heat flow relay structure, reaction relay chain, and reaction imbalance region includes: Read the heat flow relay structure, reaction relay chain, reaction imbalance region and heat accumulation region, extract the heat source output control object, heat exchange channel control object, valve opening control object and cooling medium flow control object, and generate a set of multi-source heat regulation objects. Based on the position of the reaction imbalance region in the reaction relay chain, identify the corresponding region's insufficient heat supply state, heat accumulation state, and heat transfer interruption state, and configure the heat input, heat output, or heat transfer control type for each reaction imbalance region. According to the connection sequence of the heat source initiation area, heat flow receiving area, reaction imbalance area and heat recovery area in the heat flow relay structure, determine the start-up sequence, output level and action time of each heat source output control object, and generate heat source output adjustment command. According to the heat source output regulation command, the heat exchange channel control object, valve opening control object and cooling medium flow control object are synchronously adjusted so that heat enters the reaction imbalance area with insufficient heat supply along the heat flow relay structure, and the excess heat in the heat accumulation area is directed to the heat recovery area. During the execution of multi-source thermal coupling regulation, the temperature change, pressure response, and hydrogen absorption / desorption rate response of each reaction unit are monitored according to the transmission sequence of the reaction relay chain. The reaction units that have reached the reaction propagation conditions are marked as the units that have been accepted, while the reaction units that have not reached the reaction propagation conditions are retained as the units to be regulated. The set of multi-source thermal regulation objects is updated according to the units to be regulated.

[0014] Optionally, the closed-loop thermal control process includes: After the multi-source thermal coupling control is executed, temperature data, pressure data, hydrogen absorption and desorption rate data, heat source output status, heat exchange channel status, valve opening status and cooling medium flow status of each reaction unit are collected, and control response data are generated according to time slice and spatial location. By comparing the control response data with the multi-source thermal state data set before control, the reaction units that have completed the reaction and taken over, the reaction units that are still in the reaction-limited state, the heat accumulation and dissipation area, and the reaction imbalance residual area are identified, and control feedback records are generated. Based on the regulation feedback record, the multi-source thermal state data set is updated, the thermal potential mirror field and reaction relay field are reconstructed, the thermal path unit corresponding to the reaction imbalance residual region is marked as the re-regulation object, the thermal path unit corresponding to the heat accumulation and dissipation region is marked as the maintenance object, and the heat flow relay structure is updated accordingly. The updated thermal potential mirror field is input into the wavelet neural operator model to predict the thermal potential migration in the next time window. Based on the updated reaction relay field and heat flow relay structure, the next round of multi-source thermal coupling control commands is generated, and the control commands are returned to the heat source output control object, heat exchange channel control object, valve opening control object, and cooling medium flow control object to form a closed-loop thermal control process.

[0015] A multi-source thermally coupled solid-state hydrogen storage thermal control system according to an embodiment of the present invention includes: The data construction module is used to collect thermal coupling state data and perform preprocessing to construct a multi-source thermal state data set; The thermal potential prediction module is used to construct a thermal potential mirror field based on a multi-source thermal state data set, and to generate thermal potential migration trajectories and thermal accumulation regions by improving the wavelet neural operator model. The reaction relay module is used to construct a reaction relay field based on the thermal potential migration trajectory and the thermal accumulation region, and to generate a reaction relay chain and a reaction imbalance region. The thermal path reorganization module is used to reorganize the thermal path relationships based on the reaction relay chain and the reaction imbalance region to form a heat flow relay structure. The heat control module is used to perform multi-source heat coupling control based on the heat flow relay structure, so that heat is directionally transferred to the reaction imbalance region. The closed-loop update module is used to collect operational data after regulation, update the multi-source thermal state data set, and reconstruct the thermal potential mirror field, reaction relay field, and heat flow relay structure.

[0016] Optionally, the computer-readable storage medium stores computer program instructions, which, when executed by a processor, perform a multi-source thermally coupled solid-state hydrogen storage thermal control method.

[0017] The beneficial effects of this invention are: This invention collects thermally coupled state data from a solid-state hydrogen storage system and constructs a multi-source thermal state data set, enabling temperature distribution, pressure response, hydrogen absorption and desorption rates, heat source input, and cooling status to be correlated and processed under a unified time and space reference. This overcomes the limitations of existing technologies that rely solely on local temperature feedback for regulation and improves the completeness and accuracy of thermal state perception within the hydrogen storage bed.

[0018] This invention constructs a thermal potential mirror field and introduces an improved wavelet neural operator model to identify the thermal diffusion inheritance relationship, thermal potential competition relationship, and thermal potential migration direction. It can generate thermal potential migration trajectory and heat accumulation area within a future time window, enabling the system to predict local overheating, insufficient heating, and abnormal heat migration locations in advance. This improves the foresight of thermal control and reduces thermal shock and control fluctuations caused by hysteresis adjustment.

[0019] This invention links the hydrogen storage reaction propulsion state with multi-source heat flow paths by constructing a reaction relay field, a reaction relay chain, and a heat flow relay structure. This enables heat to be directionally transferred to the reaction imbalance region along the reorganized heat path, achieving coordinated regulation of heat source output, heat exchange intensity, and cooling medium flow rate, thereby improving the uniformity of hydrogen storage reaction propulsion, hydrogen absorption and desorption efficiency, and system thermal stability. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a multi-source thermally coupled solid-state hydrogen storage thermal control method proposed in this invention; Figure 2 This is a training loss variation curve of the improved wavelet neural operator model for a multi-source thermally coupled solid-state hydrogen storage thermal control method proposed in this invention. Figure 3 This is a comparison diagram of the predicted thermal potential migration trajectory and the actual thermal potential change of a multi-source thermally coupled solid-state hydrogen storage thermal control method proposed in this invention. Figure 4 This is a schematic diagram of the heat flow relay structure of a multi-source thermally coupled solid-state hydrogen storage thermal control method proposed in this invention; Figure 5 This is a schematic diagram of the structure of a multi-source thermally coupled solid-state hydrogen storage thermal control system proposed in this invention. Detailed Implementation

[0021] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0022] refer to Figures 1-4 A multi-source thermally coupled solid-state hydrogen storage thermal control method, comprising: Collect thermal coupling state data from a solid-state hydrogen storage system, preprocess the thermal coupling state data, and construct a multi-source thermal state data set; A thermal potential mirror field is constructed based on a multi-source thermal state data set. The thermal potential mirror field is input into an improved wavelet neural operator model to identify the thermal diffusion inheritance relationship, thermal potential competition relationship and thermal potential migration direction, and generate the thermal potential migration trajectory and thermal accumulation area within the future time window. Based on thermal potential migration trajectory and thermal accumulation region, a reaction relay field is constructed to identify the reaction transmission relationship, reaction blocking relationship and reaction competition relationship between different regions of the hydrogen storage bed, and to generate reaction relay chain and reaction imbalance region. Based on the reaction relay chain and reaction imbalance region, a heat path association structure is constructed to identify the heat flow inheritance relationship, heat flow competition relationship and heat flow blocking relationship among multiple heat sources. The heat path association relationship is reorganized according to the transmission order of the reaction relay chain to form a heat flow relay structure. Based on the heat flow relay structure, reaction relay chain and reaction imbalance region, multi-source thermal coupling control is performed to coordinate the heat output, heat exchange intensity and cooling medium flow rate of each heat source, so that heat is directionally transferred to the reaction imbalance region along the heat flow relay structure. Real-time acquisition of temperature distribution data, pressure data, and hydrogen absorption / desorption rate data after regulation; updating of multi-source thermal state data set; and reconstruction of thermal potential mirror field, reaction relay field, and heat flow relay structure to form a closed-loop thermal control process.

[0023] In this embodiment, the thermal coupling state data includes temperature distribution data, pressure data, hydrogen absorption and desorption rate data, heat source input data, and cooling system operation data.

[0024] In this embodiment, the preprocessing of the thermally coupled state data to construct a multi-source thermal state data set includes: Collect temperature data, pressure data, hydrogen absorption and desorption rate data, heat source input data, and cooling system operation data at different locations in the hydrogen storage bed, and generate corresponding time markers according to the collection time; Time alignment processing is performed on thermally coupled state data with different sampling periods, and a correspondence is established based on a unified time identifier to generate a synchronous thermal state data sequence. Based on the spatial location of each sensor, a spatial coordinate mapping relationship for the hydrogen storage bed is established, and each data item in the synchronous thermal state data sequence is associated with the corresponding spatial location to generate a spatial thermal state data sequence. Abnormal, missing, and duplicate data in the spatial thermal state data sequence are corrected and combined according to time identifier and spatial location to construct a multi-source thermal state data set.

[0025] In this embodiment, generating the thermal potential migration trajectory and thermal accumulation region within the future time window includes: An improved wavelet neural operator model is constructed, which includes an input lifting layer, a two-branch wavelet transform layer, a cross-branch attention layer, an operator update layer, and an output mapping layer. The input lifting layer and the two-branch wavelet transform layer are connected by a residual connection, the two-branch wavelet transform layer and the cross-branch attention layer are connected by a parallel connection, the cross-branch attention layer and the operator update layer are connected by a serial connection, and the operator update layer and the output mapping layer are connected by a skip connection. The improved wavelet neural operator model is constructed as follows: Based on the input mapping, wavelet transform, operator update, and output mapping structure of the traditional wavelet neural operator model, this paper makes improvements. Specifically, the original input mapping structure is transformed into an input lifting layer, which is used to uniformly map the thermal potential mirror field, pressure change, and hydrogen absorption and desorption rates into high-dimensional thermal state features. A residual connection is added after the input lifting layer to preserve the original thermal potential distribution. The original single wavelet transform structure is transformed into a two-branch wavelet transform layer. The local branch is used to extract the local thermal accumulation and short-range thermal diffusion features of the hydrogen storage bed, and the global branch is used to extract the overall thermal potential migration features under the combined action of multiple heat sources. The two branches receive the high-dimensional thermal state features output by the input lifting layer in parallel. A cross-branch attention layer is added after the bi-branch wavelet transform layer to assign weights to local thermal diffusion features, global thermal potential migration features, pressure change features, and hydrogen absorption / desorption rate features. This enables the model to distinguish thermal diffusion inheritance relationships, thermal potential competition relationships, and thermal potential migration directions. The traditional operator update structure is transformed into an operator update layer with adaptive gating, which is used to adjust the update ratio of local and global features according to different reaction stages. The output mapping structure is retained and transformed into an output mapping layer. At the same time, a jump connection is set between the operator update layer and the output mapping layer, so that the output mapping layer generates the thermal potential migration trajectory and thermal accumulation region within the future time window based on the updated thermal potential acceptance characteristics, thermal potential competition characteristics and thermal potential migration direction characteristics. The model training uses historical thermal potential mirror fields and corresponding future thermal potential fields as supervision samples. The loss function is the weighted sum of prediction error and thermal potential gradient consistency error. The initial learning rate is 0.001, which decays to 90% every 10 iterations. The batch size is 16, and the number of training rounds is no less than 50. The optimizer uses the adaptive moment estimation method, and the weight decay coefficient is 0.0005. Read the thermally coupled state data, and write the temperature, heat source power, and cooling intensity at different spatial locations of the hydrogen storage bed into a three-dimensional thermal state tensor. Generate an initial thermal potential field with spatial index as the row and column dimensions and time index as the channel dimension. Specifically, the generation of the initial thermal potential field is as follows: The collected temperature, heat source power, and cooling intensity data are associated with the three-dimensional coordinate system of the hydrogen storage bed according to the sensor identifier. In each time slice, the temperature value at the same spatial location is first written into the corresponding voxel unit, then the heat source power is written into the voxel energy channel according to the power density, and the cooling intensity is written into the voxel heat dissipation channel according to the cooling flux. In the same time slice, the temperature value is normalized according to the heat source power density and cooling flux, and the missing voxels are filled by three-dimensional spatial interpolation. For voxel temperature values ​​with abnormal jumps, the average of the sliding window of adjacent time slices is used for replacement; after the three-dimensional voxel filling is completed, the voxel matrices generated in each time slice are superimposed with the spatial index of the voxel center point as the row and column dimensions and the time slice order as the channel dimension to form a thermal state tensor containing three channels: temperature, heat source power density and cooling flux. The initial thermal potential value is calculated with the temperature distribution and heat power-heat dissipation difference distribution of the thermal state tensor in the initial time slice, and the corresponding initial thermal potential field is generated. Based on the initial thermal potential field, the thermal potential gradient direction at each spatial location is calculated. The thermal potential field at the current moment is mapped and associated with the thermal potential fields at each moment within a preset time window to generate a thermal potential mirror field and record the mirror time identifier, wherein: The direction of the thermal potential gradient at each spatial location is calculated based on the initial thermal potential field, specifically as follows: Select any spatial location in the initial thermal potential field as the central voxel, retrieve the thermal potential values ​​of the six adjacent voxels above, below, left, right, front, and back of the voxel, compare the thermal potential value of the central voxel with the thermal potential values ​​of each adjacent voxel, and arrange the thermal potential values ​​from high to low to obtain the thermal potential decreasing direction. Perform the same six-axis comparison on all voxels in turn on the three orthogonal axes, record the coordinates of the adjacent voxels corresponding to the thermal potential decreasing direction of each voxel, and after recording, take the decreasing direction of the same voxel that appears most frequently on the three orthogonal axes as the main thermal potential gradient direction of the voxel, map the main thermal potential gradient direction into a direction vector and write it into the direction channel of the initial thermal potential field. After marking the main thermal potential gradient direction for all voxels, a thermal potential gradient field with vector labels is obtained, which includes temperature channel, heat source power channel, cooling flux channel and direction channel. The preset time window is 10 minutes. The thermal potential mirror field is input into the lifting layer, which uses a one-dimensional convolution operation with a kernel size of three and a channel number of one hundred and twenty-eight to map the thermal potential mirror field to a high-dimensional feature space. High-dimensional features are fed into a two-branch wavelet transform layer, where the local branch performs discrete wavelet transform to extract local multi-scale heat diffusion features, and the global branch performs continuous wavelet transform to extract global thermal potential change features. The outputs of the two branches are fused through residual connection and fed into a cross-branch attention layer. A reaction-coupled attention module is set up in the cross-branch attention layer, using hydrogen absorption / desorption rates and pressure changes as external gate control signals to dynamically adjust the weight ratio of local and global features, generating a set of thermal potential evolution features. This set of features is then fed into the operator update layer, which employs a three-layer fully connected structure with an adaptive gating unit, outputting thermal potential inheritance features, thermal potential competition features, and thermal potential migration direction features, among which: The weight ratio between local and global features is dynamically adjusted, specifically as follows: After aligning the hydrogen absorption / desorption rate and pressure change over time, normalize them into a gating vector of the same dimension. Calculate the feature mean for both local and global branch outputs. Map the gating vector to two control coefficients between zero and one. The local control coefficient increases with the increase of the hydrogen absorption / desorption rate, while the global control coefficient increases with the increase of the pressure change amplitude. Within the cross-branch attention layer, weight the local and global feature channels channel by channel based on the two control coefficients. The weight of the local feature channel is proportional to the local control coefficient, and the weight of the global feature channel is proportional to the global control coefficient. Keep the sum of the two weights constant at one. Concatenate the weighted local and global features by channel and output them through a linear mapping layer to obtain a set of thermal potential evolution features that integrate external gating information. The characteristics of output heat potential reception, heat potential competition, and heat potential migration direction are as follows: The thermal potential evolution feature set is divided into continuous segments according to time sequence. For each segment, the degree of consistency between the local thermal potential increase / decrease and the thermal potential gradient of the spatial neighborhood is calculated. Feature channels that meet the conditions of heat diffusion along the trend are summarized as thermal potential inheritance features. The thermal potential increments from different heat source directions in the same spatial region are compared. When two or more thermal potential increments tend to the same region at the same time and have similar magnitudes, the corresponding feature channels are merged and marked as thermal potential competition features. The main thermal potential gradient direction of each spatial location in the continuous time slice is vector accumulated, and the direction vector with the highest proportion of the continuous direction is extracted as the thermal potential migration direction feature. The three types of features are independently encoded in the channel dimension. The inheritance feature is used to describe the ability of heat to flow along the trend, the competition feature is used to describe the trend of heat meeting from multiple heat sources, and the migration direction feature is used to describe the overall direction of heat flow. The thermal potential reception characteristics, thermal potential competition characteristics, and thermal potential migration direction characteristics are input into the output mapping layer. The output mapping layer uses a deconvolution structure to generate the thermal potential field at each moment within the future time window. These fields are then connected in chronological order to generate the thermal potential migration trajectory. The locations in the thermal potential migration trajectory where the accumulated thermal potential exceeds a preset aggregation threshold are identified as thermal aggregation regions. Generate the thermal potential field at each moment within the future time window, specifically as follows: In the output mapping layer, the thermal potential inheritance features, thermal potential competition features, and thermal potential migration direction features are first input into a three-layer cascaded deconvolution unit in the form of channel splicing. The first layer deconvolution unit restores the spatial resolution of the thermal potential features with spatial deconvolution with a stride of 2. The second layer deconvolution unit expands the single-frame feature sequence into multiple consecutive time frame features with temporal deconvolution with a stride of 2. The third layer deconvolution unit maps the expanded time frame features into a thermal potential field consistent with the spatial size of the hydrogen storage bed with channel deconvolution with a stride of 1. Multiple consecutive thermal potential fields are spliced ​​together according to the time index order to obtain a thermal potential field sequence covering a preset time window, that is, the thermal potential field at each moment in the future time window. The default preset aggregation threshold is set to 1.2 times the average global thermal potential of the hydrogen storage bed within the same time slice.

[0026] In this embodiment, the generation of the reaction relay chain and the reaction imbalance region includes: The thermal potential migration trajectory, thermal accumulation region, pressure data, and hydrogen absorption / desorption rate data are read. The hydrogen storage bed is divided into multiple reaction units according to its spatial location. For each reaction unit, the thermal potential arrival time, thermal potential residence state, thermal potential source region, pressure response state, and hydrogen absorption / desorption rate response state are recorded. The hydrogen storage bed is divided into multiple reaction units according to its spatial location, specifically: When dividing the reaction unit according to the spatial location of the hydrogen storage bed, the distribution of heat exchange channels, the arrangement of heat sources and the location of temperature sensors are first used as the basis. The hydrogen storage bed is divided into several equidistant or equal-volume slices in the X, Y and Z axes respectively. The slices are overlapped to form a grid voxel in three-dimensional space. The side length of the voxel is set to twice the temperature measurement coverage radius of the sensor or the equivalent thermal conduction radius of the heat exchange channel, so that each voxel contains at least one temperature sampling point or is adjacent to at least one heat exchange pipe. Next, adjacent voxels that are spatially continuous and have similar thermal coupling conditions (consistent temperature gradient direction, same heat source type, and similar heat conduction paths) are merged to obtain an irregular region with consistent thermal conduction characteristics. This region is defined as a reaction unit. If the volume after merging is still greater than the preset upper limit of 250 cm³, the reaction unit is considered a reaction unit. 3 Then, based on the volume threshold, further subdivision is carried out to eventually form a set of reaction units covering the entire volume of the hydrogen storage bed. Each reaction unit has a unique spatial index, which corresponds one-to-one with the thermal potential arrival time, pressure response state, and hydrogen absorption / desorption rate response state in the upper control logic. Based on the thermal potential arrival time, thermal potential residence state, pressure response state and hydrogen absorption / desorption rate response state of each reaction unit, the reaction advancement state of the corresponding reaction unit is determined. The reaction unit that satisfies the continuous arrival of thermal potential, synchronous pressure response and effective change of hydrogen absorption / desorption rate is marked as the reaction receiving unit. The reaction unit that lags in pressure response or has insufficient change in hydrogen absorption / desorption rate after the arrival of thermal potential is marked as the reaction restricted unit. According to the migration direction of the thermal potential migration trajectory, the reaction advancement status between adjacent reaction units is compared sequentially. The adjacent relationship where the previous reaction unit has completed the reaction advancement and the next reaction unit can continue to take over the reaction advancement is marked as a reaction transfer relationship. The adjacent relationship where the next reaction unit cannot take over the reaction advancement of the previous reaction unit is marked as a reaction blocking relationship. The adjacent relationship where multiple reaction units compete for heat in the same thermal accumulation area in the same time slice is marked as a reaction competition relationship. By combining reaction units, reaction propagation states, reaction transmission relationships, reaction blocking relationships, and reaction competition relationships according to time slices and spatial locations, a reaction relay field is constructed. Reaction units with reaction transmission relationships are sequentially connected along the direction of thermal potential migration to generate a reaction relay chain. Specifically, the construction of the reaction relay field involves: Within each time slice, a reaction state matrix is ​​generated with spatial index as the horizontal axis and reaction unit identifier as the vertical axis. The reaction propagation state of each reaction unit in the corresponding time slice is written into the matrix unit. Within the same time slice, directional identifiers are inserted into the matrix for adjacent reaction units with reaction transmission relationship, blocking identifiers are inserted into the matrix for adjacent reaction units with reaction blocking relationship, and competition identifiers are inserted into the matrix for adjacent reaction units with reaction competition relationship. The reaction state matrices of each time slice are superimposed in chronological order to form a three-dimensional grid with time as the vertical axis and space as the horizontal axis. In the three-dimensional grid, with reaction units as nodes, adjacent nodes with reaction transmission relationship in the same time slice are connected by directional edges. The reaction propagation state from the same node in consecutive time slices is connected by vertical edges. The reaction transmission direction, blocking identifier, or competition identifier is recorded on the edge. Continuous directional edges are retrieved along the thermal potential migration direction in the three-dimensional grid and connected to the reaction units they point to in sequence to form a reaction relay chain. The node containing the blocking identifier or competition identifier is used as the entry mark of the reaction imbalance region to complete the construction of the reaction relay field. The reaction interruption and competition relationships in the reaction relay field are read. The reaction unit corresponding to the interruption position of the reaction relay chain is identified as the reaction discontinuity unit. The reaction unit in the thermal accumulation region whose reaction propagation state has not reached the set reaction propagation conditions is identified as the thermal reaction mismatch unit. The spatial ranges of the reaction discontinuity unit and the thermal reaction mismatch unit are merged to generate the reaction imbalance region, in which: The reaction propulsion conditions are as follows: within the target time slice, the local temperature of the hydrogen storage bed is within ±5℃ of the material's optimal reaction temperature range, and the local temperature change rate is not less than 0.2℃·min-1; at the same time, the hydrogen absorption and desorption rate at this location reaches more than 80% of the corresponding material's maximum calibrated absorption and desorption rate, and the pressure fluctuation range is kept not exceeding 3kPa. A reaction unit that meets the above three indicators of temperature, rate, and pressure is considered to have achieved reaction propulsion. If any one of the indicators is not met at the same time, it is considered that the reaction propulsion conditions have not been met. The region of reaction imbalance is generated, specifically: The positions of the discontinuous reaction units and the thermal reaction mismatch units are mapped in the three-dimensional spatial coordinate system. Voxels that overlap or are in contact with each other are merged according to their spatial adjacency to form initial imbalance blocks. Each initial imbalance block is expanded by eight neighborhoods. Voxels that share faces, edges or vertices with each other and have not met the reaction advancement conditions in two consecutive time slices are merged into the corresponding imbalance blocks to obtain continuous imbalance regions. Continuous imbalance regions are filtered according to volume thresholds. Imbalance regions with volumes less than the threshold are marked as negligible fragments. Imbalance regions with volumes not less than the threshold are determined as reaction imbalance regions. Their spatial boundaries, center coordinates and the index of the reaction relay chain to which they belong are recorded.

[0027] In this embodiment, the step of reorganizing the thermal path relationships according to the transmission sequence of the reaction relay chain to form a heat flow relay structure includes: Read the reaction relay chain, reaction imbalance region, thermal potential migration trajectory, heat accumulation region and heat source layout data, establish thermal path units according to the hydrogen storage bed spatial partition, and write the heat source source, starting region, target region, heat exchange channel, heat transfer medium, cooling medium, valve status, heat transfer direction and controllable execution object for each thermal path unit; According to the transmission sequence of the reaction relay chain, the positional relationship and heat transfer direction between each heat path unit and the reaction unit are compared. The heat path unit that can connect the previous reaction unit and the next reaction unit is marked as the heat flow receiving unit. The heat path unit that has multiple heat sources acting on the same reaction imbalance region is marked as the heat flow competing unit. The heat path unit whose heat transfer direction is inconsistent with the transmission sequence of the reaction relay chain or cannot reach the next reaction unit is marked as the heat flow blocking unit. The heat flow receiving units, heat flow competing units, and heat flow blocking units are associated according to spatial adjacency, temporal sequence, and heat source relationship to construct a heat path association structure. This structure records the receiving target, competing object, blocking position, and adjustable control object corresponding to each heat path unit. Specifically, the construction of the heat path association structure is as follows: Within the same time slice, heat flow receiving units, heat flow competing units, and heat flow blocking units that are in contact with each other or share heat exchange interfaces are merged into local heat path segments according to their spatial adjacency. According to the temporal order, local heat path segments that overlap in the starting area and the target area in adjacent time slices are longitudinally connected to obtain longitudinal heat path segments that reflect the continuous direction of heat transfer. Based on the relationship of heat source, the longitudinal heat path segments driven by the same heat source or acting on the same heat source control valve group are merged laterally to form the main heat path corresponding to the heat source. During the merging process, each main heat path is marked with its receiving target, competing object, blocking position and adjustable control object. All main heat paths are combined in the spatial coordinate system according to the starting region, receiving target and main control heat source to obtain a complete heat path association structure; Based on the location of the reaction imbalance region and the transmission order of the reaction relay chain, the heat path association structure is reorganized. The heat flow receiving unit pointing to the reaction imbalance region is retained, the output order of the heat source corresponding to the heat flow competition unit is adjusted, the heat flow blocking unit that causes heat deviation is closed or weakened, and the alternative heat path unit that can reach the next reaction unit is matched for the heat flow blocking position. The reorganized heat path units are connected in the order of heat source initiation region, heat flow receiving region, reaction imbalance region, and heat recovery region to form a heat flow relay structure. The heat source output control object, heat exchange channel control object, valve opening control object, and cooling medium flow control object corresponding to the heat flow relay structure are output. Specifically, the heat flow relay structure is formed as follows: The reorganized heat path units are grouped, with the starting area where the same heat source output end is located as the first node. The connected heat flow receiving area, the reaction imbalance area traversed, and the heat recovery area to which the final flow is directed are connected in series into a continuous path according to the transmission order. The multiple parallel paths corresponding to the same heat source starting area are prioritized, and the path that can directly reach the reaction imbalance area and has a complete heat recovery path is given priority. The remaining paths are arranged in order according to the principle of decreasing distance and risk of blockage. The heat exchange channels or valve openings shared between adjacent paths are merged to ensure that the same control object appears only once in the heat flow relay structure and its position in the entire path is marked. Then, the paths from different heat sources but flowing to the same reaction imbalance area are connected in sequence according to the merging order. At the merging node, the competition relationship and the adjustable valve object are marked. All the paths connected in sequence are recorded in the spatial coordinate system as a heat flow relay structure composed of the heat source starting area, heat flow receiving area, reaction imbalance area and heat recovery area in sequence. The output power setpoint of all heat sources, the start and stop signals of heat exchange channels, the target value of valve opening and the target value of cooling medium flow are extracted on the path to generate the corresponding heat source output control object, heat exchange channel control object, valve opening control object and cooling medium flow control object.

[0028] In this embodiment, the multi-source thermal coupling control based on the heat flow relay structure, reaction relay chain, and reaction imbalance region includes: Read the heat flow relay structure, reaction relay chain, reaction imbalance region and heat accumulation region, extract the heat source output control object, heat exchange channel control object, valve opening control object and cooling medium flow control object, and generate a set of multi-source heat regulation objects. Based on the position of the reaction imbalance region in the reaction relay chain, identify the corresponding region's insufficient heat supply state, heat accumulation state, and heat transfer interruption state, and configure the heat input, heat output, or heat transfer control type for each reaction imbalance region. According to the connection sequence of the heat source initiation area, heat flow receiving area, reaction imbalance area and heat recovery area in the heat flow relay structure, determine the start-up sequence, output level and action time of each heat source output control object, and generate heat source output adjustment command. According to the heat source output regulation command, the heat exchange channel control object, valve opening control object and cooling medium flow control object are synchronously adjusted so that heat enters the reaction imbalance area with insufficient heat supply along the heat flow relay structure, and the excess heat in the heat accumulation area is directed to the heat recovery area. During the execution of multi-source thermal coupling regulation, the temperature change, pressure response, and hydrogen absorption / desorption rate response of each reaction unit are monitored according to the transmission sequence of the reaction relay chain. The reaction units that have reached the reaction propagation conditions are marked as the units that have been accepted, while the reaction units that have not reached the reaction propagation conditions are retained as the units to be regulated. The set of multi-source thermal regulation objects is updated according to the units to be regulated.

[0029] In this embodiment, the closed-loop thermal control process includes: After the multi-source thermal coupling control is executed, temperature data, pressure data, hydrogen absorption and desorption rate data, heat source output status, heat exchange channel status, valve opening status and cooling medium flow status of each reaction unit are collected, and control response data are generated according to time slice and spatial location. By comparing the control response data with the multi-source thermal state data set before control, the reaction units that have completed the reaction and taken over, the reaction units that are still in the reaction-limited state, the heat accumulation and dissipation area, and the reaction imbalance residual area are identified, and control feedback records are generated. Based on the regulation feedback record, the multi-source thermal state data set is updated, the thermal potential mirror field and reaction relay field are reconstructed, the thermal path unit corresponding to the reaction imbalance residual region is marked as the re-regulation object, the thermal path unit corresponding to the heat accumulation and dissipation region is marked as the maintenance object, and the heat flow relay structure is updated accordingly. The updated thermal potential mirror field is input into the wavelet neural operator model to predict the thermal potential migration in the next time window. Based on the updated reaction relay field and heat flow relay structure, the next round of multi-source thermal coupling control commands is generated, and the control commands are returned to the heat source output control object, heat exchange channel control object, valve opening control object, and cooling medium flow control object to form a closed-loop thermal control process.

[0030] refer to Figure 5 A multi-source thermally coupled solid-state hydrogen storage thermal control system, comprising: The data construction module is used to collect thermal coupling state data and perform preprocessing to construct a multi-source thermal state data set; The thermal potential prediction module is used to construct a thermal potential mirror field based on a multi-source thermal state data set, and to generate thermal potential migration trajectories and thermal accumulation regions by improving the wavelet neural operator model. The reaction relay module is used to construct a reaction relay field based on the thermal potential migration trajectory and the thermal accumulation region, and to generate a reaction relay chain and a reaction imbalance region. The thermal path reorganization module is used to reorganize the thermal path relationships based on the reaction relay chain and the reaction imbalance region to form a heat flow relay structure. The heat control module is used to perform multi-source heat coupling control based on the heat flow relay structure, so that heat is directionally transferred to the reaction imbalance region. The closed-loop update module is used to collect operational data after regulation, update the multi-source thermal state data set, and reconstruct the thermal potential mirror field, reaction relay field, and heat flow relay structure.

[0031] In this embodiment, the computer-readable storage medium stores computer program instructions, which, when executed by a processor, perform a multi-source thermally coupled solid-state hydrogen storage thermal control method.

[0032] Example 1: In a continuous hydrogen absorption simulation cycle, the system receives a batch of thermally coupled state data from a solid-state hydrogen storage bed. The effective volume of the hydrogen storage bed is 8.4L, filled with 18kg of metal hydride hydrogen storage material. Internally, it has 48 temperature sampling points, 6 pressure sampling points, and 2 hydrogen flow rate sampling points. Externally, it is equipped with 4 groups of zoned electric heating units, each with a rated power of 250W. Internally, it has 2 cooling medium channels with an initial cooling flow rate of 1.8L / min. Before hydrogen absorption begins, the average temperature of the hydrogen storage bed is 32.4℃, the maximum temperature difference is 3.1℃, the inlet pressure is 1.6MPa, and the initial hydrogen absorption rate is 0.18g / s.

[0033] After the data enters the processing flow, the system continuously collects temperature, pressure, hydrogen absorption rate, heat source power, and cooling flow rate at 10-second sampling intervals, obtaining a total of 360 time slices. Since the pressure sampling period is 5 seconds, the temperature sampling period is 10 seconds, and the flow rate sampling period is 2 seconds, the system first aligns the data according to a uniform 10-second time slice. For missing temperature points, it uses joint interpolation of adjacent spatial points and adjacent time slices. Before processing, the temperature data contained 17 anomalous jump points, with the largest jump amplitude being 8.6℃. After sliding window correction, the number of anomalous jump points was reduced to 2, and the standard deviation of the temperature series decreased from 4.92 to 3.37.

[0034] The system divides the hydrogen storage bed space into 12×8×6 voxels, recording the temperature in the temperature channel, the heat source power in the heat source channel according to power density, and the cooling flow rate in the cooling flux channel, forming a thermal state tensor with dimensions of 12×8×6×3. At the 8th minute, the temperature in the upper region at the inlet was 51.8℃, in the middle region was 43.5℃, and in the lower region at the far end was 37.9℃. The heat source channel showed that the heating power on the inlet side was still 60W, and the cooling channel showed that the cooling flow rate on the inlet side was 1.8L / min. The system generates an initial thermal potential field based on the difference between the heat source power and the cooling flux, with the thermal potential value of the upper inlet layer being 0.82, the average thermal potential value of the entire bed being 0.61, and the thermal potential value of the lower region at the far end being 0.39.

[0035] During the thermal potential mirror field construction phase, the system uses the current thermal potential field as a reference to perform temporal mapping and spatial alignment of the thermal potential fields from the previous 10 minutes (60 frames) to generate a thermal potential mirror field. The system retrieves the thermal potential values ​​of the six adjacent voxels (upper, lower, left, right, front, and rear) for each voxel to obtain the main thermal potential gradient direction. At the 8th minute, 73% of the voxels in the upper inlet region have their main direction pointing towards the upper-middle region, indicating that heat is migrating from the inlet side into the bed. In the lower-middle region, 61% of the voxels are in a low thermal potential receiving state, but the corresponding hydrogen absorption rate is only 0.44 g / s, indicating insufficient heat reception in this region.

[0036] The improved wavelet neural operator model was trained using historical thermal potential mirror fields and corresponding future thermal potential fields. A total of 12,000 training samples were used, with 9,600 for training, 1,200 for validation, and 1,200 for testing. The input samples were continuous 10-minute thermal potential mirror fields, and the supervision label was the actual thermal potential field for the next 10 minutes. The model input enhancement layer uplifted the 3-channel thermal potential mirror field to 128-dimensional features. A bi-branch wavelet transform layer extracted local heat accumulation features and global thermal potential migration features, respectively. A cross-branch attention layer used hydrogen absorption rate and pressure change as gating signals. The initial learning rate was 0.001, decaying to 90% every 10 epochs. The batch size was 16, and training lasted 60 epochs with a weight decay coefficient of 0.0005. After training, the average prediction error of the thermal potential field on the test set was 4.6%, the accuracy rate for identifying heat accumulation regions was 91.8%, and the accuracy rate for identifying the thermal potential migration direction was 89.7%.

[0037] At the 10-minute mark, the model predicted that the accumulated heat potential in the upper inlet area would reach 1.26 times the average heat potential of the entire bed within the next 10 minutes, exceeding the preset accumulation threshold of 1.2 times. Therefore, this area was designated as a heat accumulation zone. In actual operation at the 18-minute mark, the temperature in the upper inlet area reached 66.4℃, a deviation of 1.5℃ from the model's prediction of 64.9℃; the actual temperature in the lower middle layer was 39.6℃, a deviation of 0.8℃ from the model's prediction of 40.4℃. This indicates that the model can anticipate local heat accumulation and insufficient heating trends.

[0038] During the reaction relay field construction phase, the system merges voxels into 42 reaction units according to spatial adjacency, with each reaction unit having a volume not exceeding 250 cm³. 3 The system records the thermal potential arrival time, thermal potential residence state, pressure response state, and hydrogen absorption rate response state for each reaction unit. The upper inlet reaction unit experiences thermal potential arrival at 12 minutes and maintains thermal potential for 6 minutes, but its hydrogen absorption rate decreases from 0.91 g / s to 0.64 g / s, and its pressure response amplitude increases from 2.8 kPa to 4.2 kPa, thus it is marked as a thermally mismatched unit. The lower middle reaction unit experiences thermal potential arrival at 19 minutes, with a hydrogen absorption rate of only 0.48 g / s, failing to meet the 80% reaction propagation condition for the maximum hydrogen absorption rate, and is marked as a discontinuous reaction unit. The system connects reaction units satisfying the reaction transfer relationship along the thermal potential migration direction, generating three reaction relay chains, with the second relay chain discontinuous in the lower middle layer. After spatial merging, the volume of the imbalance region in the upper inlet is 510 cm³. 3 The volume of the unbalanced area in the lower middle section is 760 cm³. 3 All exceeded 250cm 3 Threshold.

[0039] During the formation of the thermal path association structure, the system identified 28 thermal path units, including 13 heat flow receiving units, 7 heat flow competing units, and 8 heat flow blocking units. The inlet-side electric heating path overlapped with the heat accumulation area and was marked as a heat flow competing unit; the cooling channel near the lower middle layer maintained a flow rate of 1.8 L / min and was marked as a heat flow blocking unit; the material heat conduction path from the upper inlet layer to the lower middle layer followed the same direction as the reaction relay chain and was marked as a heat flow receiving unit. After reorganizing the thermal paths, the system formed two heat flow relay structures, one for guiding excess heat from the upper inlet layer into the recovery area, and the other for transferring heat from the middle heat source to the reaction imbalance area in the lower middle layer.

[0040] During the control and execution phase, the system reduced the inlet-side electric heating power from 60W to 15W, increased the inlet-side cooling flow rate from 1.8L / min to 2.6L / min, and increased the inlet heat exchange valve opening from 45% to 68%. Simultaneously, the middle-section electric heating power was increased from 80W to 145W, the lower middle-section cooling flow rate was reduced from 1.8L / min to 1.1L / min, and the heat exchange channel from the middle to the lower layer was opened. After 8 minutes of control, the inlet upper-section temperature decreased from 66.4℃ to 58.1℃, the lower middle-section temperature increased from 39.6℃ to 48.9℃, the hydrogen absorption rate of the inlet upper-section recovered from 0.64g / s to 0.83g / s, and the hydrogen absorption rate of the lower middle-section increased from 0.48g / s to 0.76g / s.

[0041] Under the same initial conditions, the method of this invention was compared with the traditional temperature threshold feedback method. The traditional method only increases the cooling flow rate after the local temperature exceeds 65°C, with a maximum local temperature of 69.8°C, a maximum temperature difference in the hydrogen storage bed of 21.5°C, a thermal accumulation duration of 18.6 min, an average hydrogen absorption rate of 0.69 g / s, a time required to reach the target hydrogen storage capacity of 74 min, and a total cooling medium consumption of 168 L. The method of this invention has a maximum local temperature of 61.7°C, a maximum temperature difference of 9.4°C, a thermal accumulation duration of 5.2 min, an average hydrogen absorption rate of 0.82 g / s, a time required to reach the target hydrogen storage capacity of 59 min, and a total cooling medium consumption of 142 L. The traditional method had 19 reaction units that did not meet the reaction propagation conditions, while the method of this invention had only 6. The hydrogen storage utilization rate of the traditional method was 91.3%, while that of the method of this invention was 97.4%.

[0042] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A multi-source thermally coupled solid-state hydrogen storage thermal control method, characterized in that, include: Collect thermal coupling state data from a solid-state hydrogen storage system, preprocess the thermal coupling state data, and construct a multi-source thermal state data set; A thermal potential mirror field is constructed based on a multi-source thermal state data set. The thermal potential mirror field is input into an improved wavelet neural operator model to identify the thermal diffusion inheritance relationship, thermal potential competition relationship and thermal potential migration direction, and generate the thermal potential migration trajectory and thermal accumulation area within the future time window. Based on thermal potential migration trajectory and thermal accumulation region, a reaction relay field is constructed to identify the reaction transmission relationship, reaction blocking relationship and reaction competition relationship between different regions of the hydrogen storage bed, and to generate reaction relay chain and reaction imbalance region. Based on the reaction relay chain and reaction imbalance region, a heat path association structure is constructed to identify the heat flow inheritance relationship, heat flow competition relationship and heat flow blocking relationship among multiple heat sources. The heat path association relationship is reorganized according to the transmission order of the reaction relay chain to form a heat flow relay structure. Based on the heat flow relay structure, reaction relay chain and reaction imbalance region, multi-source thermal coupling control is performed to coordinate the heat output, heat exchange intensity and cooling medium flow rate of each heat source, so that heat is directionally transferred to the reaction imbalance region along the heat flow relay structure. Real-time acquisition of temperature distribution data, pressure data, and hydrogen absorption / desorption rate data after regulation; updating of multi-source thermal state data set; and reconstruction of thermal potential mirror field, reaction relay field, and heat flow relay structure to form a closed-loop thermal control process.

2. The multi-source thermally coupled solid-state hydrogen storage thermal control method according to claim 1, characterized in that, The thermal coupling status data includes temperature distribution data, pressure data, hydrogen absorption and desorption rate data, heat source input data, and cooling system operation data.

3. The multi-source thermally coupled solid-state hydrogen storage thermal control method according to claim 1, characterized in that, The preprocessing of the thermally coupled state data to construct a multi-source thermal state data set includes: Collect temperature data, pressure data, hydrogen absorption and desorption rate data, heat source input data, and cooling system operation data at different locations in the hydrogen storage bed, and generate corresponding time markers according to the collection time; Time alignment processing is performed on thermally coupled state data with different sampling periods, and a correspondence is established based on a unified time identifier to generate a synchronous thermal state data sequence. Based on the spatial location of each sensor, a spatial coordinate mapping relationship for the hydrogen storage bed is established, and each data item in the synchronous thermal state data sequence is associated with the corresponding spatial location to generate a spatial thermal state data sequence. Abnormal, missing, and duplicate data in the spatial thermal state data sequence are corrected and combined according to time identifier and spatial location to construct a multi-source thermal state data set.

4. The multi-source thermally coupled solid-state hydrogen storage thermal control method according to claim 1, characterized in that, The generation of the thermal potential migration trajectory and thermal accumulation region within the future time window includes: An improved wavelet neural operator model is constructed, which includes an input lifting layer, a two-branch wavelet transform layer, a cross-branch attention layer, an operator update layer, and an output mapping layer. The input lifting layer and the two-branch wavelet transform layer are connected by a residual connection, the two-branch wavelet transform layer and the cross-branch attention layer are connected by a parallel connection, the cross-branch attention layer and the operator update layer are connected by a serial connection, and the operator update layer and the output mapping layer are connected by a skip connection. Read the thermal coupling state data, write the temperature, heat source power and cooling intensity of different spatial locations of the hydrogen storage bed into the three-dimensional thermal state tensor, and generate the initial thermal potential field with the spatial index as the row and column dimension and the time index as the channel dimension. Based on the initial thermal potential field, the thermal potential gradient direction at each spatial location is calculated. The thermal potential field at the current moment is mapped and associated with the thermal potential fields at each moment within the preset time window to generate a thermal potential mirror field and record the mirror time identifier. The thermal potential mirror field is input into the lifting layer, which uses a one-dimensional convolution operation with a kernel size of three and a channel number of one hundred and twenty-eight to map the thermal potential mirror field to a high-dimensional feature space. High-dimensional features are fed into a two-branch wavelet transform layer, where the local branch performs discrete wavelet transform to extract local multi-scale heat diffusion features, and the global branch performs continuous wavelet transform to extract global thermal potential change features. The outputs of the two branches are fused through residual connection and fed into a cross-branch attention layer. A reaction-coupled attention module is set in the cross-branch attention layer, using the hydrogen absorption and desorption rate and pressure change as external gate control signals to dynamically adjust the weight ratio of local features and global features, and generate a set of thermal potential evolution features. The set of thermal potential evolution features is then sent to the operator update layer, which adopts a three-layer fully connected structure and inserts an adaptive gate control unit to output thermal potential inheritance features, thermal potential competition features, and thermal potential migration direction features. The thermal potential reception characteristics, thermal potential competition characteristics, and thermal potential migration direction characteristics are input into the output mapping layer. The output mapping layer uses a deconvolution structure to generate the thermal potential field at each moment within the future time window. The thermal potential migration trajectory is generated by connecting them in chronological order. The position in the thermal potential migration trajectory where the accumulated thermal potential exceeds the preset aggregation threshold is determined as the thermal aggregation region.

5. The multi-source thermally coupled solid-state hydrogen storage thermal control method according to claim 1, characterized in that, The generated reaction relay chain and reaction imbalance region include: Read the thermal potential migration trajectory, thermal accumulation region, pressure data and hydrogen absorption and desorption rate data, divide the hydrogen storage bed into multiple reaction units according to the spatial location of the hydrogen storage bed, and write the thermal potential arrival time, thermal potential residence state, thermal potential source region, pressure response state and hydrogen absorption and desorption rate response state for each reaction unit. Based on the thermal potential arrival time, thermal potential residence state, pressure response state and hydrogen absorption / desorption rate response state of each reaction unit, the reaction advancement state of the corresponding reaction unit is determined. The reaction unit that satisfies the continuous arrival of thermal potential, synchronous pressure response and effective change of hydrogen absorption / desorption rate is marked as the reaction receiving unit. The reaction unit that lags in pressure response or has insufficient change in hydrogen absorption / desorption rate after the arrival of thermal potential is marked as the reaction restricted unit. According to the migration direction of the thermal potential migration trajectory, the reaction advancement status between adjacent reaction units is compared sequentially. The adjacent relationship where the previous reaction unit has completed the reaction advancement and the next reaction unit can continue to take over the reaction advancement is marked as a reaction transfer relationship. The adjacent relationship where the next reaction unit cannot take over the reaction advancement of the previous reaction unit is marked as a reaction blocking relationship. The adjacent relationship where multiple reaction units compete for heat in the same thermal accumulation area in the same time slice is marked as a reaction competition relationship. By combining reaction units, reaction propagation states, reaction transmission relationships, reaction blocking relationships, and reaction competition relationships according to time slices and spatial locations, a reaction relay field is constructed, and reaction units with reaction transmission relationships are sequentially connected along the direction of thermal potential migration to generate a reaction relay chain. Read the reaction blocking relationship and reaction competition relationship in the reaction relay field, determine the reaction unit corresponding to the interruption position of the reaction relay chain as the reaction discontinuity unit, determine the reaction unit in the thermal accumulation area whose reaction propagation state has not reached the set reaction propagation conditions as the thermal reaction mismatch unit, merge the spatial range of the reaction discontinuity unit and the thermal reaction mismatch unit to generate the reaction imbalance region.

6. The multi-source thermally coupled solid-state hydrogen storage thermal control method according to claim 1, characterized in that, The reorganization of thermal path relationships according to the transmission sequence of the reaction relay chain to form a heat flow relay structure includes: Read the reaction relay chain, reaction imbalance region, thermal potential migration trajectory, heat accumulation region and heat source layout data, establish thermal path units according to the hydrogen storage bed spatial partition, and write the heat source source, starting region, target region, heat exchange channel, heat transfer medium, cooling medium, valve status, heat transfer direction and controllable execution object for each thermal path unit; According to the transmission sequence of the reaction relay chain, the positional relationship and heat transfer direction between each heat path unit and the reaction unit are compared. The heat path unit that can connect the previous reaction unit and the next reaction unit is marked as the heat flow receiving unit. The heat path unit that has multiple heat sources acting on the same reaction imbalance region is marked as the heat flow competing unit. The heat path unit whose heat transfer direction is inconsistent with the transmission sequence of the reaction relay chain or cannot reach the next reaction unit is marked as the heat flow blocking unit. The heat flow receiving unit, heat flow competing unit, and heat flow blocking unit are associated according to spatial adjacency, temporal sequence, and heat source relationship to construct a heat path association structure. The receiving target, competing object, blocking position, and adjustable control object corresponding to each heat path unit are recorded in the heat path association structure. Based on the location of the reaction imbalance region and the transmission order of the reaction relay chain, the heat path association structure is reorganized. The heat flow receiving unit pointing to the reaction imbalance region is retained, the output order of the heat source corresponding to the heat flow competition unit is adjusted, the heat flow blocking unit that causes heat deviation is closed or weakened, and the alternative heat path unit that can reach the next reaction unit is matched for the heat flow blocking position. The reorganized thermal path units are connected in the order of heat source initiation area, heat flow receiving area, reaction imbalance area and heat recovery area to form a heat flow relay structure, and output the heat source output control object, heat exchange channel control object, valve opening control object and cooling medium flow control object corresponding to the heat flow relay structure.

7. The multi-source thermally coupled solid-state hydrogen storage thermal control method according to claim 1, characterized in that, The multi-source thermal coupling control based on the heat flow relay structure, reaction relay chain, and reaction imbalance region includes: Read the heat flow relay structure, reaction relay chain, reaction imbalance region and heat accumulation region, extract the heat source output control object, heat exchange channel control object, valve opening control object and cooling medium flow control object, and generate a set of multi-source heat regulation objects. Based on the position of the reaction imbalance region in the reaction relay chain, identify the corresponding region's insufficient heat supply state, heat accumulation state, and heat transfer interruption state, and configure the heat input, heat output, or heat transfer control type for each reaction imbalance region. According to the connection sequence of the heat source initiation area, heat flow receiving area, reaction imbalance area and heat recovery area in the heat flow relay structure, determine the start-up sequence, output level and action time of each heat source output control object, and generate heat source output adjustment command. According to the heat source output regulation command, the heat exchange channel control object, valve opening control object and cooling medium flow control object are synchronously adjusted so that heat enters the reaction imbalance area with insufficient heat supply along the heat flow relay structure, and the excess heat in the heat accumulation area is directed to the heat recovery area. During the execution of multi-source thermal coupling regulation, the temperature change, pressure response, and hydrogen absorption / desorption rate response of each reaction unit are monitored according to the transmission sequence of the reaction relay chain. The reaction units that have reached the reaction propagation conditions are marked as the units that have been accepted, while the reaction units that have not reached the reaction propagation conditions are retained as the units to be regulated. The set of multi-source thermal regulation objects is updated according to the units to be regulated.

8. The multi-source thermally coupled solid-state hydrogen storage thermal control method according to claim 1, characterized in that, The closed-loop thermal control process includes: After the multi-source thermal coupling control is executed, temperature data, pressure data, hydrogen absorption and desorption rate data, heat source output status, heat exchange channel status, valve opening status and cooling medium flow status of each reaction unit are collected, and control response data are generated according to time slice and spatial location. By comparing the control response data with the multi-source thermal state data set before control, the reaction units that have completed the reaction and taken over, the reaction units that are still in the reaction-limited state, the heat accumulation and dissipation area, and the reaction imbalance residual area are identified, and control feedback records are generated. Based on the regulation feedback record, the multi-source thermal state data set is updated, the thermal potential mirror field and reaction relay field are reconstructed, the thermal path unit corresponding to the reaction imbalance residual region is marked as the re-regulation object, the thermal path unit corresponding to the heat accumulation and dissipation region is marked as the maintenance object, and the heat flow relay structure is updated accordingly. The updated thermal potential mirror field is input into the wavelet neural operator model to predict the thermal potential migration in the next time window. Based on the updated reaction relay field and heat flow relay structure, the next round of multi-source thermal coupling control commands are generated, and the control commands are returned to the heat source output control object, heat exchange channel control object, valve opening control object and cooling medium flow control object to form a closed-loop thermal control process.

9. A multi-source thermally coupled solid-state hydrogen storage thermal control system, executing the multi-source thermally coupled solid-state hydrogen storage thermal control method according to any one of claims 1 to 8, characterized in that, include: The data construction module is used to collect thermal coupling state data and perform preprocessing to construct a multi-source thermal state data set; The thermal potential prediction module is used to construct a thermal potential mirror field based on a multi-source thermal state data set, and to generate thermal potential migration trajectories and thermal accumulation regions by improving the wavelet neural operator model. The reaction relay module is used to construct a reaction relay field based on the thermal potential migration trajectory and thermal accumulation region, and generate reaction relay chain and reaction imbalance region. The thermal path reorganization module is used to reorganize the thermal path relationships based on the reaction relay chain and the reaction imbalance region to form a heat flow relay structure; The heat control module is used to perform multi-source heat coupling control based on the heat flow relay structure, so that heat is directionally transferred to the reaction imbalance region. The closed-loop update module is used to collect operational data after regulation, update the multi-source thermal state data set, and reconstruct the thermal potential mirror field, reaction relay field, and heat flow relay structure.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed by a processor, perform a multi-source thermally coupled solid-state hydrogen storage thermal control method as described in any one of claims 1-8.

Images