Power supply power consumption control method and system for low-power wearable intelligent glasses

Abstract

The invention discloses a power supply power consumption control method and system for low-power wearable intelligent glasses, and relates to the technical field of wearable equipment power management, and the method comprises the following steps: obtaining role priority data and a space-sharing coupling matrix of each target object; wherein the space-sharing coupling matrix represents the sharing capability of the target object on different functional modules at each spatial position; and performing weighted fusion on the role priority data and the space-sharing coupling matrix, performing directional neighborhood correction on a weighted fusion result according to a cross-mode identification set, generating a module availability matrix, performing local decomposition and aggregation alignment on the module availability matrix, and generating a cross factor set. According to the scheme, the energy consumption distribution relation between the functional modules can be corrected in real time according to the spatial position difference and the cross factor set constraint through the iterative solution process, so that the sensitivity and convergence stability of the priority matrix to the system operation state change are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of power management technology for wearable devices, specifically to a power consumption control method and system for low-power wearable smart glasses. Background Technology

[0002] With the development of augmented reality, mixed reality, and human-machine collaboration technologies, wearable smart glasses are gradually becoming key terminals in various collaborative task scenarios such as on-site operations, rescue, inspection, teaching, and remote collaboration. These wearable devices often need to integrate multiple functional modules such as cameras, displays, positioning, communication, sensing, and edge computing. At the same time, in collaborative scenarios, multiple wearing devices often need to share data in real time and complement each other's functions to meet the timeliness and reliability requirements of the task. However, due to limitations in battery capacity, device power consumption, and wearing comfort, battery life has always been the core bottleneck affecting the practicality and deployment scale of smart glasses. Therefore, in multi-user collaborative task scenarios, how to perform global power and wake-up scheduling for multiple devices and modules while ensuring task quality and responsiveness has become an important technical problem in this field.

[0003] Existing technologies mainly assign fixed priorities to different roles and adjust processor frequencies or process computing tasks in layers according to task load. The advantage of these solutions is that they can finely control computing power consumption and extend device battery life to a certain extent. However, resource scheduling based solely on initial priorities may cause critical modules to be frequently woken up in actual low-priority environments, resulting in unnecessary power consumption increases. In addition, static priorities lack quantitative consideration of sharing capabilities and are difficult to effectively utilize the collaborative advantages of highly shared functional modules in space. This may lead to multiple devices repeatedly performing the same high-power operations, further wasting power. Summary of the Invention

[0004] The purpose of this invention is to provide a power consumption control method and system for low-power wearable smart glasses to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows:

[0006] In a first aspect, the present invention discloses a power consumption control method for low-power wearable smart glasses, applied to power consumption control of wearable smart glasses in collaborative task scenarios, comprising the following steps:

[0007] Obtain the role priority data and spatial-shared coupling matrix of each target object; whereby the spatial-shared coupling matrix represents the target object's ability to share different functional modules at each spatial location;

[0008] The role priority data and the space-sharing coupling matrix are weighted and fused. The weighted fusion result is then corrected in a directional neighborhood according to the cross-module identifier set to generate a module availability matrix. The module availability matrix is ​​then locally decomposed and aggregated to generate a cross factor set.

[0009] The cross-modal identifier set is generated by spatial clustering and cross-generating the role priority data and the space-sharing coupling matrix.

[0010] Using the cross factor set as constraints, the module availability matrix and the role priority data are iteratively solved according to spatial location until a preset number of iterations is reached, generating a dynamic priority matrix.

[0011] The dynamic priority matrix is ​​constructed hierarchically according to the functional module dimension to generate a threshold vector set. Combined with the module availability matrix and the cross-module identifier set, a spatial clustering constraint scheduling problem is constructed and solved to generate a power allocation matrix.

[0012] Secondly, this invention discloses a power consumption control system for low-power wearable smart glasses, comprising:

[0013] The data acquisition module is used to acquire the role priority data and space-shared coupling matrix of each target object;

[0014] The cross-factor analysis module is used to perform weighted fusion of the role priority data and the space-shared coupling matrix, perform directional neighborhood correction on the weighted fusion result according to the cross-module identifier set, generate a module availability matrix, and perform local decomposition and aggregation alignment on the module availability matrix to generate a cross-factor set.

[0015] The priority division module is used to iteratively solve the module availability matrix and the role priority data with the cross factor set as constraints until a preset number of iterations is reached, thereby generating a dynamic priority matrix.

[0016] The power allocation matrix generation module is used to construct a hierarchical threshold structure for the dynamic priority matrix according to the functional module dimension, generate a threshold vector set, and combine the module availability matrix and the cross-module identifier set to construct and solve a spatial clustering constraint scheduling problem to generate a power allocation matrix.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0018] 1. This solution, through an iterative solution process, can correct the energy consumption allocation relationship between functional modules in real time based on spatial location differences and cross factor set constraints. This significantly improves the sensitivity and convergence stability of the dynamic priority matrix to changes in system operating status. Through residual analysis and reverse update mechanism, it effectively suppresses the impact of abnormal fluctuations of a single functional module on the overall power consumption scheduling, improves the energy efficiency consistency and response accuracy of the system in complex scenarios, avoids repeated wake-ups of key functional modules or unnecessary high-power operations, and significantly reduces equipment energy waste.

[0019] 2. This solution constructs a hierarchical threshold for the dynamic priority matrix according to the functional module dimension, enabling adaptive adjustment of power consumption distribution among multifunctional modules. By calculating row and column quantile values, the impact of abnormal priorities on the overall threshold is effectively reduced, improving the stability and robustness of threshold generation. At the same time, by using the weighted superposition of statistical features and intermediate correction vectors, the threshold can dynamically reflect the real-time operating status and power consumption sensitivity of each functional module, thereby achieving precise power limiting for high-load modules and automatic relaxation for low-load modules. This allows the system to significantly reduce overall energy consumption while ensuring response performance. Attached Figure Description

[0020] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts. Wherein:

[0021] Figure 1 This is a flowchart illustrating the steps of a power consumption control method for low-power wearable smart glasses according to the present invention.

[0022] Figure 2 A flowchart illustrating the generation module availability matrix provided by this invention;

[0023] Figure 3 A schematic diagram of the process for generating a dynamic priority matrix provided by the present invention;

[0024] Figure 4 A schematic diagram of the process for generating a power allocation matrix provided by the present invention;

[0025] Figure 5 This invention provides a schematic diagram of the module functions of a power consumption control system for low-power wearable smart glasses. Detailed Implementation

[0026] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0027] Application Overview:

[0028] In multi-user collaborative task scenarios for wearable smart glasses, due to the adoption of a static priority scheduling mechanism, critical functional modules are frequently woken up even in low-priority environments. At the same time, the lack of quantitative assessment of spatial sharing capabilities leads to multiple devices repeatedly performing high-power operations, resulting in wasted power consumption. This problem directly affects the device's battery life, task quality, and responsiveness. Specifically, during task execution, the power management strategy is unable to dynamically adapt to changes in the scenario, critical modules are activated at unnecessary times, and devices in close proximity fail to effectively collaborate and share functional module resources, resulting in a reduction in overall energy efficiency.

[0029] For example, in power facility inspection tasks, multiple inspectors wear wearable smart glasses to work collaboratively. The glasses need to collect equipment status images in real time and share location information. When inspectors are scattered in different areas of the substation, the existing system schedules based on fixed role priority. All devices independently wake up high-resolution cameras to collect images. Even if adjacent personnel have covered the same equipment area and shared image data, multiple devices still repeatedly perform camera wake-up and image processing operations, causing unnecessary power consumption and affecting the continuous execution capability of inspection tasks.

[0030] If the above problems are not addressed, the waste of power consumption will exacerbate the continuous deterioration of the device's battery life. During critical mission phases, this may cause the device to enter a low-power state prematurely, interrupting real-time data sharing and functional collaboration processes, reducing the reliability and timeliness of mission execution, and ultimately making it difficult to complete collaborative tasks as expected.

[0031] After introducing the basic concept of the present invention, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0032] Example 1:

[0033] Please see Figure 1 A power consumption control method for low-power wearable smart glasses, applied to power consumption control of wearable smart glasses in collaborative task scenarios, includes the following steps:

[0034] Obtain the role priority data and spatial-shared coupling matrix of each target object; whereby the spatial-shared coupling matrix represents the target object's ability to share different functional modules at each spatial location;

[0035] The role priority data and the space-sharing coupling matrix are weighted and fused. The weighted fusion result is then corrected in a directional neighborhood according to the cross-modal identifier set to generate a module availability matrix. The module availability matrix is ​​then locally decomposed and aggregated and aligned to generate a cross factor set.

[0036] Among them, the cross-modal identifier set is generated by spatial clustering cross-generating the role priority data and the spatial-shared coupling matrix;

[0037] Using the cross factor set as a constraint, the module availability matrix and role priority data are iteratively solved according to spatial location until the preset number of iterations is reached, generating a dynamic priority matrix;

[0038] The dynamic priority matrix is ​​constructed hierarchically according to the functional module dimension to generate a set of threshold vectors. Combined with the module availability matrix and cross-module identifier set, a spatial clustering constraint scheduling problem is constructed and solved to generate a power allocation matrix.

[0039] Among them, role priority data refers to quantitative information used to reflect the importance or priority of each target object in a collaborative task scenario;

[0040] The spatial-shared coupling matrix is ​​a data structure used to characterize the ability of various target objects to share functional modules at different spatial locations;

[0041] Cross-modal identifier set refers to the set generated by cross-mapping the behavioral characteristics, role priority relationships and shared capabilities of each target object in different functional modules and spatial locations;

[0042] The weighted fusion result refers to the data structure generated after uniformly quantifying and integrating role priority data and spatial-shared coupling matrix information;

[0043] Directional neighborhood correction refers to the data processing procedure that locally adjusts the module availability matrix based on spatial clustering information of cross-modal identifier sets.

[0044] The module availability matrix is ​​a data structure that quantifies the operability and shareability of the functional modules of each target object in a collaborative task scenario.

[0045] Local decomposition and aggregation alignment refers to the process of fine-grained factor extraction and global consistency integration based on spatial clustering of the generated module availability matrix;

[0046] The cross factor set refers to a set of globally unified factors generated after local decomposition, aggregation, and alignment of the module availability matrix based on spatial clustering and cross-modal identifier set;

[0047] Spatial clustering cross refers to a data processing process that systematically integrates the spatial location, functional module utilization capabilities, and priority characteristics of different target objects by jointly analyzing role priority data and spatial-sharing coupling matrices.

[0048] Iterative solution refers to the data processing process of continuously adjusting the priority allocation of functional modules on each target object through multiple rounds of calculation in order to obtain a dynamic priority matrix;

[0049] The preset number of iterations refers to the parameter used to control the update rounds of the availability matrix and role priority data of the iterative solution module;

[0050] A dynamic priority matrix is ​​a two-dimensional matrix used to depict the real-time priority status of each target object to each functional module within a specific time window.

[0051] Hierarchical threshold construction refers to the data processing process of quantifying and classifying the task priority characteristics of the dynamic priority matrix at the functional module dimension.

[0052] The threshold vector set refers to the hierarchical partitioning result of the dynamic priority matrix in terms of functional modules;

[0053] A power allocation matrix is ​​a data structure used to accurately indicate the power consumption status of each functional module of each target object in a collaborative task scenario within each time window.

[0054] This solution achieves refined power consumption scheduling by dynamically integrating role priority and spatial sharing capabilities. Specifically, the weighted fusion of role priority data and spatial-sharing coupling matrix ensures that scheduling decisions simultaneously respond to task criticality and spatial collaboration potential. Directional neighborhood correction enables the module availability matrix to accurately capture the sharing dynamics in the actual environment. The generation of cross-factor sets provides an effective constraint basis for iterative solutions. Iterative optimization of the dynamic priority matrix enables scheduling to adapt to real-time scenario changes. Hierarchical threshold construction and spatial clustering constraint scheduling further optimize power consumption allocation. Thus, it effectively overcomes the power consumption waste problem caused by static scheduling strategies, avoids critical modules being frequently woken up and repeatedly executing high-power operations in low-priority environments, and extends the device's battery life while ensuring task quality and responsiveness.

[0055] The above describes a complete scheme for power consumption control of low-power wearable smart glasses. The following section describes how to obtain the role priority data and space-shared coupling matrix of each target object, specifically including:

[0056] Read the target object list from the sensor identification registry. For each target object, extract information including but not limited to role category, permission level and historical priority score from the historical task scheduling record. Calculate the weighted average of role category and historical priority score to obtain the role priority data of the corresponding target object.

[0057] The spatial location information of the target object is obtained through GPS, and the access records of shared resources between the target objects are obtained through historical scheduling logs. The access records of shared resources include, but are not limited to, shared tasks, access frequency of shared resources, shared devices or shared communication link data. For each target object, the distance is calculated based on the spatial location coordinates, and then weighted by the access frequency of shared resources in the access records to form the spatial-sharing coupling coefficient of the target object. All spatial-sharing coupling coefficients are integrated to generate a spatial-sharing coupling matrix.

[0058] The role priority data and the space-sharing coupling matrix are weighted and fused element by element to generate a preliminary availability score matrix. The preliminary availability score matrix is ​​then used as the weighted fusion result. The specific calculation formula is as follows:

[0059] ;

[0060] In the formula, Indicates the target object in the preliminary usability score matrix For functional modules Preliminary usability score, Represents the target object Role priority data, Represents the target object For functional modules Space-shared coupling coefficient, and These represent the corresponding weighted fusion coefficients. All the above data have been normalized during the calculation.

[0061] The above describes how to obtain the role priority data and spatial-shared coupling matrix of each target object. The following describes how to perform directional neighborhood correction on the weighted fusion result based on the cross-modal identifier set to generate a module availability matrix. Please refer to [link / reference]. Figure 2 , Figure 2 This is a flowchart illustrating the process of generating a module availability matrix according to an embodiment of this application. Generating the module availability matrix specifically includes:

[0062] Based on the spatial clustering of cross-modal identifier sets, the weighted fusion results are spatially corrected to generate a correction matrix.

[0063] The correction matrix is ​​subjected to eigenvalue decomposition and spectral radius calculation. It is determined whether the spectral radius calculation result is greater than the preset discrimination threshold. If so, the eigenvalue decomposition result is scaled and corrected according to the ratio of the preset discrimination threshold and the spectral radius calculation result. The scaling and correction result is then subjected to sparse orthogonal basis projection to obtain the module availability matrix.

[0064] Among them, the correction matrix refers to the result generated after spatial neighborhood correction of the preliminary weighted fusion result;

[0065] Eigenvalue decomposition is a data processing procedure that uses linear algebraic analysis on a modified matrix to extract its internal structural features.

[0066] Spectral radius calculation refers to the data processing process of performing matrix eigenvalue analysis on the correction matrix and calculating the maximum value of the modulus of its eigenvalues.

[0067] The preset discrimination threshold refers to the upper limit of the matrix eigenvalues ​​used to constrain the module availability matrix correction process. It is obtained by collecting historical correction matrix samples, calculating the spectral radius of each historical correction matrix sample, and taking the quantile value (e.g., 75%) of the spectral radius calculation result corresponding to each historical correction matrix sample.

[0068] Sparse orthogonal basis projection is a linear algebraic processing method that constrains the scaling correction results.

[0069] The above content will be described in detail below:

[0070] Spatial clustering (such as K-means or hierarchical clustering) is performed on the role priority data and spatial-shared coupling matrix in both spatial and shared feature dimensions. Specifically, the Euclidean distance is calculated based on the spatial location of the target object to form the initial spatial cluster partition. During the clustering process, a cross-coupling strategy is applied: that is, within the same spatial cluster, the preliminary availability score corresponding to the role priority data and spatial-shared coupling coefficient is read, and it is determined whether the preliminary availability score is greater than the preset clustering threshold of the corresponding spatial cluster. Otherwise, the functional module corresponding to the target object is moved into the adjacent spatial cluster, and a cross-module identifier set is output (each element is a triple of <target object, functional module, spatial cluster>).

[0071] The clustering threshold preset for the corresponding spatial cluster is obtained by calculating the mean and standard deviation of the preliminary availability scores of the target objects contained in the corresponding spatial cluster, and subtracting a certain multiple (e.g., 2 times) from the mean of the preliminary availability scores.

[0072] Based on the spatial clustering of the cross-modal identifier set, the weighted fusion result is spatially corrected to generate a correction matrix. The specific calculation formula is as follows:

[0073] ;

[0074] In the formula, Represents the target object in the correction matrix The value, Indicates the target object in the weighted fusion result The value, Represents the target object The neighborhood set, Represents the target object Neighborhood object data, Indicates the target object in the weighted fusion result The value, This represents the neighborhood correction coefficient. All the above data have been normalized during the calculation.

[0075] The modified matrix is ​​decomposed using standard linear algebra methods (such as eigenvalue decomposition or singular value decomposition SVD) to obtain a set of eigenvalues ​​and a set of eigenvectors. The set of eigenvectors is used as the result of the eigenvalue decomposition. The maximum value of the absolute value of the eigenvalue set is then taken to obtain the spectral radius calculation result.

[0076] Determine whether the calculated spectral radius is greater than a preset discrimination threshold; if so, then evaluate the eigenvalue decomposition result. According to the preset discrimination threshold Calculation results of spectral radius The ratio is scaled and corrected, and the scaling correction result is then subjected to sparse orthogonal basis projection to obtain the module availability matrix. The specific calculation formula is as follows:

[0077] ;

[0078] In the formula, This represents the sparse orthogonal basis projection function. All the data above have been normalized during the calculation.

[0079] This scheme performs spatial neighborhood correction on the weighted fusion results based on the cross-modal identifier set to generate a correction matrix. This ensures that the correction process can accurately match the distribution characteristics of the shared capabilities of the target object at different locations. By performing eigenvalue decomposition and spectral radius calculation on the correction matrix, the iterative convergence of the matrix is ​​dynamically monitored through the spectral radius calculation results. Furthermore, the dynamic scaling mechanism based on the actual spectral radius calculation results can adaptively compress unstable feature components. This effectively solves the problems of numerical instability and sparsity loss in the generation process of the module availability matrix, ensuring that the module availability matrix can accurately reflect the distribution law of spatial shared capabilities.

[0080] The above describes the directional neighborhood correction of the weighted fusion result based on the cross-modal identifier set to generate a module availability matrix. The following describes the local decomposition and aggregation alignment of the module availability matrix to generate a cross-factor set, specifically including:

[0081] Based on the spatial clustering of cross-modal identifier sets, extract the corresponding sub-matrices from the module availability matrix;

[0082] Perform multi-scale local sparse decomposition on each submatrix to generate corresponding local factor sets, and perform consistency alignment on each local factor set according to scale differences. Perform multi-level sparsification aggregation on each consistent aligned local factor set to generate cross factor sets.

[0083] Among them, the cross-modal identifier set is generated by spatial clustering and cross-generating the role priority data and the spatial-shared coupling matrix.

[0084] Here, a submatrix refers to a local data block extracted from the module availability matrix;

[0085] Multi-scale local sparse decomposition refers to the data processing process of simultaneously performing sparse non-negative matrix decomposition on each sub-matrix in the module availability matrix, which is determined by cross-module identifier set space clustering, according to different scale levels.

[0086] The local factor set refers to the output of the multi-scale local sparse decomposition of the module availability matrix submatrix within each spatial cluster;

[0087] Consistency alignment refers to the data processing process of standardizing and unifying the coordinates of local factor sets within each spatial cluster on a global scale;

[0088] Multi-level sparsity aggregation refers to the data processing process of integrating the local factor sets after local factor consistency alignment with layer-by-layer sparse constraints to form a unified cross factor set.

[0089] The above content will be described in detail below:

[0090] Read the <target object, functional module> information of each spatial cluster in the cross-modal identifier set, and then match the corresponding target object and functional module in the module availability matrix by row and column index. Extract the module availability elements involved in each spatial cluster one by one to form a sub-matrix.

[0091] For each submatrix, divide it into sub-blocks according to a predefined scale (such as functional module category or power consumption range). Then, calculate the corresponding sub-blocks in each sub-block using non-negative matrix factorization combined with local sparsity constraints. Local factor matrix and residual matrix The specific calculation formula is as follows:

[0092] ;

[0093] In the formula, Indicates a nonnegativity constraint. Indicates a sub-block The corresponding submatrix, Represents the sparsity regularization coefficient. The square operation represents the Frobenius norm. This represents the factor matrix variables in the calculation process. Represents the residual matrix variables during the calculation process. The variable representing the first element in the factor matrix. Line number Column elements, This represents the summation of the absolute values ​​of all elements in the factor matrix variable. All the data above has been normalized during the calculation.

[0094] By integrating the local factor matrices and residual matrices corresponding to each sub-block, the local factor set of the corresponding sub-matrix is ​​obtained;

[0095] The similarity between local factor sets is calculated based on scale differences, and the specific calculation formula is as follows:

[0096] ;

[0097] In the formula, Indicates the first The local factor set and the first Similarity between local factor sets Indicates the first A local factor set, Indicates the first A local factor set, Indicates the first The local factor set and the first Scale differences between local factor sets This represents the Frobenius norm; all the data above have been normalized during the calculation.

[0098] Based on similarity, each local factor set is corrected and aligned as a whole to obtain a consistent aligned local factor set.

[0099] Sparse constraints are applied to each local factor set after consistency alignment at the row and column levels. Specifically, threshold truncation or L1 regularization is applied to remove elements that do not meet the requirements. Then, the processed local factor sets are aggregated to generate cross factor sets.

[0100] This scheme extracts corresponding sub-matrices from the module availability matrix based on the spatial clustering structure of the cross-module identifier set, realizing the segmentation of global data into local regions. Multi-scale local sparse decomposition is performed on each sub-matrix to generate corresponding local factor sets, capturing shared capability features from micro to macro levels. Then, each local factor set is aligned consistently based on scale differences to eliminate feature distortion caused by scale jumps. Finally, multi-level sparse aggregation processing is performed on the aligned local factor sets to generate cross-factor sets. Through the above scheme, this approach can generate more accurate cross-factor sets, effectively improving the quality of subsequent iterative solutions for the dynamic priority matrix.

[0101] The above describes the local decomposition and aggregation alignment of the module availability matrix to generate a cross-factor set. The following describes using the cross-factor set as a constraint to iteratively solve the module availability matrix and role priority data according to spatial location until a preset number of iterations is reached, generating a dynamic priority matrix. Please refer to [reference needed]. Figure 3 , Figure 3 This is a flowchart illustrating the process of generating a dynamic priority matrix according to an embodiment of this application. Generating the dynamic priority matrix specifically includes:

[0102] The module availability matrix and role priority data are weighted and calculated to generate an initial priority matrix;

[0103] In each round of iterative solution, the cross factor set is used as a constraint term, and the initial priority matrix is ​​subjected to residual analysis according to its spatial location. The initial priority matrix is ​​then updated in reverse based on the results of the residual analysis.

[0104] Once the preset number of iterations is reached, the residual analysis results generated during each iteration are accumulated to generate an intermediate correction vector, and the updated initial priority matrix generated during the current iteration is used as the dynamic priority matrix.

[0105] The initial priority matrix refers to the basic priority allocation table for each functional module in wearable smart glasses.

[0106] Residual analysis refers to the data processing process that quantifies the spatial deviation between the module availability matrix and the initial priority matrix in each iteration.

[0107] The intermediate correction vector is a data structure used to adjust priority matrix information during quantization and accumulation iteration.

[0108] The above content will be described in detail below:

[0109] The module availability matrix and role priority data are weighted and calculated to generate an initial priority matrix. The specific calculation formula is as follows:

[0110] ;

[0111] In the formula, Represents the target object in the initial priority matrix For functional modules elements, Represents the target object Role priority data, Represents the target object in the module availability matrix For functional modules elements, and These represent the corresponding weighted calculation coefficients. All the above data have been normalized during the calculation.

[0112] In each round of iterative solution, the initial priority matrix is ​​divided into several initial priority sub-matrices according to spatial location. Then, the cross factor set is mapped to each initial priority sub-matrix. The residual between each element of the initial priority sub-matrix and the cross factor set is calculated by difference. All residuals are combined to obtain the residual analysis result. Based on the positive and negative signs and magnitude information of each residual in the residual analysis result, the priority of the corresponding elements in the initial priority sub-matrix is ​​adjusted according to the preset reverse update rule. The update results of each initial priority sub-matrix are merged back into the overall initial priority matrix to obtain the updated initial priority matrix.

[0113] Once the preset number of iterations is reached, the residual analysis results generated during each iteration are accumulated to generate an intermediate correction vector, and the updated initial priority matrix generated during the current iteration is used as the dynamic priority matrix.

[0114] This solution, through an iterative solution process, can correct the energy consumption allocation relationship between functional modules in real time based on spatial location differences and cross factor set constraints. This significantly improves the sensitivity and convergence stability of the priority matrix to changes in system operating status. Through residual analysis and reverse update mechanism, it effectively suppresses the impact of abnormal fluctuations of a single functional module on the overall power consumption scheduling, improves the energy efficiency consistency and response accuracy of the system in complex scenarios, avoids repeated wake-ups of key modules or unnecessary high-power operations, and significantly reduces equipment energy waste.

[0115] The above describes the iterative solution of the module availability matrix and role priority data based on spatial location, using the cross-factor set as a constraint, until a preset number of iterations is reached, generating a dynamic priority matrix. The following describes the hierarchical threshold construction of the dynamic priority matrix according to the functional module dimension, generating a threshold vector set, specifically including:

[0116] The quantile values ​​of the dynamic priority matrix are calculated independently from both row and column perspectives, and the dynamic priority matrix is ​​updated based on the quantile value calculation results.

[0117] The updated dynamic priority matrix is ​​subjected to feature statistics at the functional module level, and the statistical feature results and intermediate correction vectors are weighted and superimposed to generate the threshold vector set of the corresponding functional module.

[0118] Among them, quantile value calculation refers to the statistical processing of each row and each column of the dynamic priority matrix;

[0119] Statistical characteristic results refer to the quantitative indicators used to describe the priority distribution and fluctuation characteristics of each functional module, calculated on the functional module dimension of the updated dynamic priority matrix.

[0120] The above content will be described in detail below:

[0121] The quantile values ​​of the dynamic priority matrix are calculated independently from both row and column perspectives. The following explanation uses the row perspective as an example:

[0122] For each row of elements, count the elements and independently calculate the first quartile (25th quartile) and the third quartile (75th quartile). Subtract the first quartile from the third quartile to obtain the interquartile range.

[0123] Determine whether the element in the row is greater than the threshold set for the row; if so, the element is determined to be an abnormal element.

[0124] Determine whether the element in the row is less than the lower threshold of the threshold range set for the row; if so, the element is determined to be an abnormal element.

[0125] The threshold for the threshold range set for this row is obtained by adding a certain multiple (e.g., 0.75 times) of the interquartile range to the third quartile.

[0126] The lower threshold of the threshold range set for this row is obtained by subtracting the interquartile range by a certain multiple (e.g., 0.5 times) from the first quartile;

[0127] The same judgment is performed on columns as above. If an element is judged to be abnormal only at the row level, if it is determined to be less than the lower threshold of the threshold range set for that row, the abnormal element is replaced with the lower threshold of the threshold range set for that row. If it is determined to be greater than the upper threshold of the threshold range set for that row, the abnormal element is replaced with the upper threshold of the threshold range set for that row.

[0128] If an element is determined to be abnormal in both rows and columns, and it is determined to be less than both the lower threshold of the threshold range set for the row and the lower threshold of the threshold range set for the column, the abnormal element is replaced with the larger of the two threshold ranges. If it is determined to be greater than both the upper threshold of the threshold range set for the row and the upper threshold of the threshold range set for the column, the abnormal element is replaced with the smaller of the two threshold ranges. Otherwise, the abnormal element is removed.

[0129] The dynamic priority matrix is ​​updated based on the quantile value calculation results;

[0130] The intermediate correction vectors are normalized, and the updated dynamic priority matrix is ​​statistically analyzed along the functional module dimension, including calculating the mean and standard deviation of each functional module. Then, combined with the normalized intermediate correction vectors, a distribution-based adaptive thresholding method is used to construct a threshold vector set for each functional module, containing three levels of thresholds.

[0131] The first threshold is obtained by adding a certain multiple (e.g., 0.5 times) of the standard deviation of the functional modules to the mean of the functional modules and a normalized intermediate correction vector of a certain multiple (e.g., 0.5 times).

[0132] The second threshold is obtained by adding a certain multiple (e.g., 0.45 times) of the standard deviation of the functional modules to the mean of the functional modules;

[0133] The third threshold is obtained by subtracting the standard deviation of the functional modules by a certain factor (e.g., 0.25 times) from the mean of the functional modules.

[0134] This solution constructs a hierarchical threshold structure for the dynamic priority matrix according to the functional module dimension, enabling adaptive adjustment of power consumption distribution among multifunctional modules. By calculating row and column quantile values, the impact of abnormal priorities on the overall threshold is effectively reduced, improving the stability and robustness of threshold generation. At the same time, by utilizing the weighted superposition of statistical features and intermediate correction vectors, the threshold can dynamically reflect the real-time operating status and power consumption sensitivity of each functional module, thereby achieving precise power limiting for high-load modules and automatic relaxation for low-load modules. This allows the system to significantly reduce overall energy consumption while ensuring response performance.

[0135] The above describes the hierarchical threshold construction of the dynamic priority matrix according to the functional module dimension, generating a threshold vector set. The following describes and combines the module availability matrix and cross-module identifier set to construct and solve a spatial clustering constraint scheduling problem, generating a power allocation matrix. Please refer to [reference needed]. Figure 4 , Figure 4 This is a schematic diagram of the process for generating a power allocation matrix provided in an embodiment of this application. Generating the power allocation matrix specifically includes:

[0136] The updated dynamic priority matrix is ​​classified according to the module availability matrix based on the threshold vector set to generate an initial wake-up set;

[0137] The initial wake-up set is mapped and retrieved based on the spatial clustering of the cross-mode identifier set, and the wake-up sequence is generated by solving the time scheduling problem based on the updated dynamic priority matrix.

[0138] The wake-up sequence, the updated dynamic priority matrix, and the intermediate correction vector are subjected to conflict rollback verification to generate a power allocation matrix.

[0139] The initial wake-up set refers to the set of potential wake-up candidate states for each functional module in the wearable smart glasses.

[0140] Mapping retrieval refers to the process of spatial clustering, localization, and attribution confirmation of each target object-functional module combination in the initial wake-up set using the acquired cross-modal identifier set;

[0141] Time scheduling solution refers to the optimization process of determining the specific wake-up time and power consumption level of each target object-functional module combination within a specified time window for the initial wake-up set after spatial clustering mapping;

[0142] The wake-up sequence refers to the scheduling and control sequence generated by the system for each functional module in the time dimension based on the initial wake-up set, cross-module identifier set, module availability matrix, and updated dynamic priority matrix.

[0143] Conflict rollback verification refers to the entire process of verifying the consistency and constraints of the wake-up sequence, dynamic priority matrix, and intermediate correction vector during power consumption control.

[0144] The above content will be described in detail below:

[0145] Based on the module availability matrix, construct a target object-functional module combination that represents the availability of the target object to the functional module. For each target object-functional module combination, extract the corresponding element from the updated dynamic priority matrix and compare it with the threshold vector set.

[0146] If the corresponding element is greater than the first threshold, the corresponding element is marked as high priority; if it is between the second threshold and the first threshold, the corresponding element is marked as medium priority; if it is between the third threshold and the second threshold, the corresponding element is marked as low priority.

[0147] The base score is calculated based on the comparison results and the threshold vector set. The specific calculation formula is as follows:

[0148] ;

[0149] In the formula, Represents the target object With functional modules The base score, This represents the overflow scale constant. Represents the target object in the updated dynamic priority matrix. For functional modules elements, Indicates functional modules The first threshold, Indicates functional modules The second threshold, Indicates functional modules The third threshold; all the above data have been normalized during the calculation.

[0150] The space-shared coupling matrix is ​​subjected to projection constraints to obtain the constraint matrix, and its specific calculation formula is as follows:

[0151] ;

[0152] In the formula, Represents the target object in the constraint matrix For functional modules elements, This represents the mapping function; all the data above have been normalized during the calculation.

[0153] The base score is weighted and calculated using the corresponding elements of the constraint matrix:

[0154] Subtract the corresponding element of the constraint matrix from 1 to obtain the reliability factor of the corresponding element, and multiply the reliability factor by the base score of the corresponding element to obtain the weighted calculation result;

[0155] If the weighted calculation result is greater than the preset constraint threshold, then the corresponding target object-functional module combination is integrated to generate a preliminary wake-up set.

[0156] Based on the spatial clustering of the cross-modal identifier set, a mapping retrieval is performed on each target object and functional module in the initial wake-up set to determine the relationship between each target object-functional module combination and its corresponding spatial cluster. This relationship is then used to partition and map the module availability matrix. Specifically, during the partitioning and mapping process, the module availability matrix is ​​remapped based on the spatial clustering information. The energy consumption of each functional module within its corresponding spatial cluster is calculated, generating the final energy consumption mapping table. The specific calculation formula is as follows:

[0157] ;

[0158] In the formula, Spatial clustering in the energy consumption mapping table Medium functional modules Energy consumption value, Spatial clustering A collection of all functional modules in the program. Represents the target object in the module availability matrix For functional modules elements, Represents the target object in the module availability matrix For functional modules The elements above have all been normalized during the calculations;

[0159] Based on the spatial clustering of the cross-modal identifier set, the initial wake-up set and energy consumption mapping table are mapped to the functional module set within each spatial cluster. Next, the constraint matrix and threshold vector set of the corresponding functional module within the spatial cluster are read, and the constraint matrix is ​​normalized according to the functional module dimension to generate the constraint weight matrix within the spatial cluster.

[0160] The energy consumption mapping table, constraint weight matrix, and threshold vector set are combined to form a constrained scheduling objective function, which is then solved to generate the optimal clustering solution. The specific calculation formula is as follows:

[0161]

[0162] ;

[0163] In the formula, Represents the set of feasible regions with constraints. Represents the candidate scheduling matrix. Spatial clustering in the candidate scheduling matrix Corresponding functional modules The allocation flag, 0 indicates unallocated, 1 indicates allocated. Represents the total number of spatial clusters. Indicates the total number of functional modules. and These represent the corresponding weight coefficients. This represents a threshold-based penalty function; all the data above have been normalized during the calculation.

[0164] The optimal clustering solution and the updated dynamic priority matrix are discretized by time window and power consumption levels are allocated to generate a wake-up sequence.

[0165] Read the allocation entries in the optimal clustering solution one by one (each entry specifies the allocation of functional modules within a spatial cluster), and calculate the scheduling time-domain parameters of each spatial cluster based on the local time stability characteristics of the updated dynamic priority matrix (obtained by calculating the local mean and local standard deviation of the updated dynamic priority matrix). The specific calculation formula is as follows:

[0166] ;

[0167] In the formula, Indicates the basic time slot length. Spatial clustering The local mean of the updated dynamic priority matrix. Spatial clustering The local standard deviation of the updated dynamic priority matrix. All the above data have been normalized during the calculation.

[0168] Based on the scheduling time-domain parameters, the corresponding space is clustered and discretized into several time windows. The specific calculation formula is as follows:

[0169] ;

[0170] In the formula, Spatial clustering Inner A time window;

[0171] Within each spatial cluster, the optimal clustering solution is sorted from high to low according to the priority of the updated dynamic priority matrix. The optimal clustering solution is then mapped to the nearest available time window according to the sorting. For multiple occupancy of the same functional module within the same time window, the lower priority entry is postponed to the next available time window within the spatial cluster based on the updated dynamic priority matrix.

[0172] After determining the allocated time window, the optimal clustering solution is mapped to the discrete power consumption level using the updated dynamic priority matrix. The specific calculation formula is as follows:

[0173] ;

[0174] In the formula, Represents the target object For functional modules In the time window Internally allocated discrete power consumption levels Represents the target object in the optimal solution of clustering. For functional modules Task allocation status, This represents the set of system configuration parameters. This represents the power consumption mapping function. All the above data have been normalized during the calculation.

[0175] Integrate time windows and discrete power consumption levels to generate a wake-up sequence (each element is a quadruple of <target object, functional module, time window, discrete power consumption level>).

[0176] The shared deployment mapping is obtained by combining the optimal clustering solution and the module availability matrix;

[0177] The system reads each allocation entry from the optimal clustering solution and performs availability checks on each entry based on the corresponding functional module's element in the module availability matrix. The availability checks include, but are not limited to, whether the remaining capacity, current load percentage, shareability, and concurrency of the functional module meet the allocation requirements for that functional module. Then, all allocation entries from the optimal clustering solutions that meet the availability checks are aggregated by functional module dimension. Combined with the corresponding elements in the module availability matrix, the module-level shared occupancy value (such as concurrency occupancy or shared bandwidth consumption) is obtained through normalization and weighted average calculation. The system determines whether the module-level shared occupancy value is greater than the preset module availability threshold. If so, the corresponding functional module, allocation entry, and corresponding element in the module availability matrix are integrated to generate a shared deployment item. All shared deployment items are summarized to generate a shared deployment mapping.

[0178] The wake-up sequence, shared deployment mapping, and module availability matrix are fused according to functional modules to generate a candidate control matrix (each element is <spatial location, target object, functional module, time window, allocation entry, corresponding element in the module availability matrix, discrete power consumption level).

[0179] The candidate control matrix is ​​searched line by line. If they are within the same spatial range (e.g., based on the target object), When there are two or more allocation entries contending for the same functional module within the same time window (e.g., within a radius of 5 meters from the center), or within the same time window (e.g., based on the target object), If two target objects within a radius of 5 meters compete for the camera module within the same time window, normalization and weighted average calculations are performed on the corresponding elements in the intermediate correction vector and the updated dynamic priority matrix to obtain the final priority value of the corresponding allocation entry. By retaining only the maximum value, it is ensured that only one allocation entry uses the same functional module within the same spatial range and the same time window. Then, the candidate control matrix is ​​adjusted to generate the power allocation matrix.

[0180] This solution optimizes the wake-up and power scheduling of low-power wearable smart glasses by combining the module availability matrix and cross-module identifier set through spatial clustering constraints. It achieves accurate classification and time scheduling of dynamic priorities for each functional module, enabling efficient allocation of the initial wake-up set within spatial clusters and reasonable scheduling of the wake-up sequence. Furthermore, it ensures the uniqueness and consistency of the power allocation matrix through conflict rollback verification. This significantly reduces overall energy consumption, improves system power utilization efficiency and scheduling reliability while ensuring the response and coverage of critical tasks. It also enhances the stability and real-time performance of multi-module collaborative work.

[0181] The above describes how to construct and solve a spatial clustering-constrained scheduling problem by combining the module availability matrix and cross-module identifier set, generating a power allocation matrix. The following describes how to classify the updated dynamic priority matrix according to the module availability matrix based on a threshold vector set, generating a preliminary wake-up set, specifically including:

[0182] Construct target object-functional module combinations based on the module availability matrix, and for each target object-functional module combination, compare the threshold vector set with the updated dynamic priority matrix, and calculate the base score based on the comparison results and the threshold vector set.

[0183] The base score and the constraint matrix are weighted and calculated. It is determined whether the weighted calculation result is greater than the preset constraint threshold. If so, the corresponding target object-functional module combination is integrated to generate a preliminary wake-up set.

[0184] The constraint matrix is ​​obtained by projecting constraints onto the space-shared coupling matrix.

[0185] Among them, the target object-functional module combination refers to the combination unit formed by mapping each functional module in the wearable smart glasses to its corresponding target object or device state.

[0186] The comparison result refers to the value obtained by comparing the priority value of each target object-functional module combination in the updated dynamic priority matrix with the threshold of the corresponding threshold vector set item by item for each combination.

[0187] The base score is a metric used to quantify the degree of match between the wake-up priority and availability of each target object-functional module combination under the current scheduling conditions;

[0188] A constraint matrix is ​​a data structure used to regulate and limit the wake-up behavior of various target objects and functional modules.

[0189] The preset constraint threshold is the boundary value used to determine whether the combination of target object and functional module meets the wake-up condition. It is obtained by calculating the statistical characteristics of the space-shared coupling matrix, such as the mean and standard deviation, and weighting the mean and standard deviation according to the pre-configured system parameters.

[0190] Projection constraint processing is a data processing technique that maps and numerically normalizes the sharing capability and location reliability of each element in a space-shared coupling matrix.

[0191] This part has already been described in detail above, so I will not repeat it here.

[0192] This solution combines a dynamic priority matrix with a set of threshold vectors and a module availability matrix, and uses a constraint matrix for weighted integration. This enables precise screening and priority wake-up control of the target object-functional module combination. As a result, while ensuring timely response of key functions, it effectively suppresses invalid wake-up of low-priority or high-risk modules, significantly reduces the overall energy consumption of the system, and improves the power utilization efficiency and response stability of wearable smart glasses in complex usage scenarios.

[0193] The above describes classifying the updated dynamic priority matrix according to the module availability matrix based on the threshold vector set to generate a preliminary wake-up set. The following describes mapping and retrieving the preliminary wake-up set based on spatial clustering of cross-module identifier sets, and then solving the time scheduling problem based on the updated dynamic priority matrix to generate a wake-up sequence. Specifically, this includes:

[0194] Based on the spatial clustering of the cross-module identifier set, the initial wake-up set is mapped and retrieved, and the module availability matrix is ​​partitioned and mapped according to the mapping retrieval results to generate an energy consumption mapping table.

[0195] Based on the energy consumption mapping table, constraint matrix, threshold vector set and mapping retrieval results, a constrained scheduling problem is constructed in each spatial cluster and the optimal solution of the cluster is solved.

[0196] The optimal clustering solution and the updated dynamic priority matrix are discretized by time window and power consumption level allocation to generate a wake-up sequence.

[0197] Partition mapping refers to the data processing process of systematically dividing and mapping the capability information of each functional module in the module availability matrix according to the spatial clustering results of the cross-module identifier set;

[0198] An energy consumption mapping table is a data structure used to describe the correspondence between the power consumption characteristics and available capabilities of each functional module at different operating levels.

[0199] Clustering optimal solution refers to the optimal scheduling result obtained within spatial clustering based on cross-modal identifier sets, considering the multi-objective balance between module wake-up, energy consumption allocation, and constraints.

[0200] Time window discretization and power consumption level allocation refers to the data processing process of dividing the continuous time axis into several adjacent non-overlapping discrete time windows based on the execution start and end times of each functional module task and the system task rhythm parameters in the optimal clustering solution, and allocating the operating power consumption level of each functional module within the discretized time windows according to the energy consumption demand vector and module availability matrix of each functional module provided by the optimal clustering solution.

[0201] This part has already been described in detail above, so I will not repeat it here.

[0202] This solution maps and retrieves the initial wake-up set based on spatial clustering of cross-modal identifiers and solves the time scheduling problem using a dynamic priority matrix. It can accurately identify the wake-up priority order of each functional module in different spatial clusters. It generates a module energy consumption mapping table through partition mapping and constructs a constrained scheduling problem in each spatial cluster by combining constraint matrix and threshold vector set and solves the optimal solution for each cluster. This achieves the minimization of energy consumption within spatial clusters and the efficient utilization of shared resources, ensuring efficient power management and dynamic energy scheduling of low-power wearable smart glasses in multi-user, multi-module collaborative environments, enhancing device battery life and optimizing real-time performance.

[0203] The above describes the mapping and retrieval of the initial wake-up set based on spatial clustering of the cross-mode identifier set, and the time scheduling solution of the mapping and retrieval results based on the updated dynamic priority matrix to generate the wake-up sequence. The following describes the conflict rollback verification of the wake-up sequence, the updated dynamic priority matrix, and the intermediate correction vector to generate the power allocation matrix, specifically including:

[0204] The wake-up sequence, shared deployment mapping, and module availability matrix are weighted and fused to generate a candidate control matrix;

[0205] The shared deployment mapping is obtained by combining the optimal clustering solution and the module availability matrix.

[0206] A weighted consensus decision is performed on the candidate control matrix, intermediate correction vector, and updated dynamic priority matrix to generate a power allocation matrix.

[0207] Among them, the shared deployment mapping refers to the data structure used to characterize the resource sharing relationship, task collaboration relationship and power consumption allocation rules between various functional modules;

[0208] The candidate control matrix is ​​a data structure used to describe the power consumption state that each functional module may take in different time windows and the corresponding cost evaluation under a given wake-up sequence and resource constraints.

[0209] Weighted consensus decision refers to the process of generating the final control instruction order and power consumption allocation through systematic calculation and weighted rules for wake-up instructions that may conflict or compete for resources in the candidate control matrix.

[0210] The decision output matrix is ​​a matrix generated after weighted consensus decision-making on the candidate control matrices, used to represent the final power consumption allocation state of each target object-functional module combination in each time window;

[0211] Consistency verification refers to the data processing process of systematically verifying and constraining the decision output matrix, module availability matrix, and shared deployment mapping.

[0212] This part has already been described in detail above, so I will not repeat it here.

[0213] This solution achieves accurate generation and multi-dimensional consistency verification of candidate control matrices by systematically integrating and weighting wake-up sequences, shared deployment mappings, module availability matrices, intermediate correction vectors, and dynamic priority matrices. This enables the effective avoidance of control command conflicts and shared resource overload in complex scheduling environments with multiple modules, users, and time windows. At the same time, weighted consistency decision-making ensures that key modules respond according to priority while taking energy efficiency optimization into account. Thus, while maintaining real-time response capabilities, it achieves low power consumption, high task coverage, and high efficiency and reliability of multi-module collaborative work.

[0214] Example 2:

[0215] Please see Figure 5 A power consumption control system for low-power wearable smart glasses, comprising:

[0216] The data acquisition module is used to acquire the role priority data and space-shared coupling matrix of each target object;

[0217] The cross-factor analysis module is used to perform weighted fusion of role priority data and spatial-shared coupling matrix, perform directional neighborhood correction on the weighted fusion result based on cross-module identifier set, generate module availability matrix, and perform local decomposition and aggregation alignment on module availability matrix to generate cross-factor set;

[0218] The priority partitioning module is used to iteratively solve the module availability matrix and role priority data with the cross factor set as constraints until the preset number of iterations is reached, and generate a dynamic priority matrix.

[0219] The power allocation matrix generation module is used to construct a hierarchical threshold for the dynamic priority matrix according to the functional module dimension, generate a threshold vector set, and combine it with the module availability matrix and cross-module identifier set to construct and solve the spatial clustering constraint scheduling problem, thereby generating the power allocation matrix.

[0220] This embodiment has the same technical effects as Embodiment 1.

[0221] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. The data mentioned in this application, when used for calculations, have undergone normalization and other preprocessing to achieve dimensional uniformity.

[0222] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A power consumption control method for low-power wearable smart glasses, applied to power consumption control of wearable smart glasses in collaborative task scenarios, characterized in that, Includes the following steps: Obtain the role priority data and spatial-shared coupling matrix of each target object; whereby the spatial-shared coupling matrix represents the target object's ability to share different functional modules at each spatial location; The role priority data and the space-sharing coupling matrix are weighted and fused. The weighted fusion result is then corrected in a directional neighborhood according to the cross-module identifier set to generate a module availability matrix. The module availability matrix is ​​then locally decomposed and aggregated to generate a cross factor set. The cross-modal identifier set is generated by spatial clustering and cross-generating the role priority data and the space-sharing coupling matrix. Using the cross factor set as constraints, the module availability matrix and the role priority data are iteratively solved according to spatial location until a preset number of iterations is reached, generating a dynamic priority matrix. The dynamic priority matrix is ​​constructed hierarchically according to the functional module dimension to generate a threshold vector set. Combined with the module availability matrix and the cross-module identifier set, a spatial clustering constraint scheduling problem is constructed and solved to generate a power allocation matrix.

2. The power consumption control method for low-power wearable smart glasses according to claim 1, characterized in that: The weighted fusion result is corrected for directional neighborhood based on the cross-modal identifier set, and the resulting module availability matrix includes: Based on the spatial clustering of the cross-modal identifier set, the weighted fusion result is spatially corrected to generate a correction matrix. The modified matrix is ​​subjected to eigenvalue decomposition and spectral radius calculation. It is determined whether the spectral radius calculation result is greater than a preset discrimination threshold. If so, the eigenvalue decomposition result is scaled and corrected according to the ratio of the preset discrimination threshold and the spectral radius calculation result. The scaling and correction result is then subjected to sparse orthogonal basis projection to obtain the module availability matrix.

3. The power consumption control method for low-power wearable smart glasses according to claim 1, characterized in that: The local decomposition and aggregation alignment of the module availability matrix to generate a cross factor set specifically includes: Based on the spatial clustering of the cross-modal identifier set, extract the corresponding sub-matrix from the module availability matrix; Perform multi-scale local sparse decomposition on each submatrix to generate corresponding local factor sets, and perform consistency alignment on each local factor set according to scale differences. Perform multi-level sparsification aggregation on each consistent aligned local factor set to generate cross factor sets.

4. The power consumption control method for low-power wearable smart glasses according to claim 1, characterized in that: Using the cross-factor set as constraints, the module availability matrix and the role priority data are iteratively solved according to spatial location until a preset number of iterations is reached, generating a dynamic priority matrix specifically includes: The module availability matrix and the role priority data are weighted and calculated to generate an initial priority matrix; In each round of iterative solution, the cross factor set is used as a constraint term, and residual analysis is performed on the initial priority matrix according to its spatial location. The initial priority matrix is ​​then updated in reverse based on the residual analysis results. Once the preset number of iterations is reached, the residual analysis results generated during each iteration are accumulated to generate an intermediate correction vector, and the updated initial priority matrix generated during the current iteration is used as the dynamic priority matrix.

5. The power consumption control method for low-power wearable smart glasses according to claim 4, characterized in that: The dynamic priority matrix is ​​constructed hierarchically according to the functional module dimension to generate a set of threshold vectors, specifically including: The quantile values ​​of the dynamic priority matrix are calculated independently from both row and column perspectives, and the dynamic priority matrix is ​​updated based on the quantile value calculation results. The updated dynamic priority matrix is ​​subjected to feature statistics at the functional module level, and the statistical feature results and the intermediate correction vector are weighted and superimposed to generate the threshold vector set of the corresponding functional module.

6. The power consumption control method for low-power wearable smart glasses according to claim 5, characterized in that: Combining the module availability matrix and the cross-module identifier set, a spatial clustering constraint scheduling problem is constructed and solved to generate a power allocation matrix, specifically including: Based on the threshold vector set, the updated dynamic priority matrix is ​​classified according to the module availability matrix to generate a preliminary wake-up set; The initial wake-up set is mapped and retrieved based on the spatial clustering of the cross-mode identifier set, and the time scheduling solution is performed on the mapping and retrieval results based on the updated dynamic priority matrix to generate a wake-up sequence. The wake-up sequence, the updated dynamic priority matrix, and the intermediate correction vector are subjected to conflict rollback verification to generate a power allocation matrix.

7. The power consumption control method for low-power wearable smart glasses according to claim 6, characterized in that: Based on the threshold vector set, the updated dynamic priority matrix is ​​classified according to the module availability matrix to generate a preliminary wake-up set, specifically including: Construct target object-functional module combinations based on the module availability matrix, and for each target object-functional module combination, compare the threshold vector set with the updated dynamic priority matrix, and calculate the base score based on the comparison result and the threshold vector set. The basic score is weighted and calculated with the constraint matrix. It is then determined whether the weighted calculation result is greater than the preset constraint threshold. If so, the corresponding target object-functional module combination is integrated to generate a preliminary wake-up set. The constraint matrix is ​​obtained by projecting constraints onto the space-shared coupling matrix.

8. The power consumption control method for low-power wearable smart glasses according to claim 7, characterized in that: The preliminary wake-up set is mapped and retrieved based on the spatial clustering of the cross-mode identifier set, and the wake-up sequence is generated by solving the time scheduling problem based on the updated dynamic priority matrix. Specifically, this includes: Based on the spatial clustering of the cross-modal identifier set, the preliminary wake-up set is mapped and retrieved, and based on the mapping retrieval results, the module availability matrix is ​​partitioned and mapped to generate an energy consumption mapping table. Based on the energy consumption mapping table, the constraint matrix, the threshold vector set, and the mapping retrieval results, a constrained scheduling problem is constructed within each spatial cluster, and the optimal solution for each cluster is solved. The optimal clustering solution and the updated dynamic priority matrix are discretized by time window and power consumption level allocation to generate a wake-up sequence.

9. The power consumption control method for low-power wearable smart glasses according to claim 8, characterized in that: The specific steps for generating a power allocation matrix include: performing conflict rollback verification on the wake-up sequence, the updated dynamic priority matrix, and the intermediate correction vector. The wake-up sequence, shared deployment mapping, and module availability matrix are weighted and fused to generate a candidate control matrix; The shared deployment mapping is obtained by combining the optimal clustering solution and the module availability matrix; A weighted consensus decision is performed on the candidate control matrix, the intermediate correction vector, and the updated dynamic priority matrix to generate a power allocation matrix.

10. A power consumption control system for low-power wearable smart glasses, characterized in that, include: The data acquisition module is used to acquire the role priority data and space-shared coupling matrix of each target object; The cross-factor analysis module is used to perform weighted fusion of the role priority data and the space-sharing coupling matrix, perform directional neighborhood correction on the weighted fusion result according to the cross-module identifier set, generate a module availability matrix, and perform local decomposition and aggregation alignment on the module availability matrix to generate a cross-factor set. The priority division module is used to iteratively solve the module availability matrix and the role priority data with the cross factor set as constraints until a preset number of iterations is reached, thereby generating a dynamic priority matrix. The power allocation matrix generation module is used to construct a hierarchical threshold structure for the dynamic priority matrix according to the functional module dimension, generate a threshold vector set, and combine the module availability matrix and the cross-module identifier set to construct and solve a spatial clustering constraint scheduling problem to generate a power allocation matrix.

Images