Deep learning-based healthy light snack market data analysis method and system

Abstract

The invention relates to the technical field of data analysis, and discloses a healthy light snack market data analysis method and system based on deep learning. The method comprises the steps of obtaining multi-field data of climbers, and performing data clustering and classification; determining a high-priority data stream according to the classified data and performing time sequence synchronization; generating a non-dislocation data chain through a dislocation detection algorithm; fusing the zero food class preference and the interactive object type to generate a balanced load index; and effect network nodes are extracted according to the balanced load indexes, a data flow control scene is simulated, and resource allocation is optimized. According to the method, the problem of resource dynamic allocation under multi-source data integration and time sequence dislocation is solved, snack consumption modes and social behaviors of climbers can be accurately analyzed under a multi-level effect structure, the data processing efficiency is optimized, and the resource scheduling accuracy and the data processing efficiency in a high-altitude environment are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data analysis, in particular to a health light snack market data analysis method and system based on deep learning. BACKGROUND

[0002] High-altitude mountaineering is a challenge to the human body limit, and its health management and behavior analysis are crucial to the safety of mountaineers and the improvement of experience. With the popularity of outdoor sports, the demand for personalized supply strategies for mountaineers is growing, especially in the management and optimization of snack consumption behavior, which not only concerns nutrition supplementation, but also involves social interaction and resource allocation efficiency. However, existing technologies often fail to meet the real-time needs in dynamic environments when dealing with this complex scenario, revealing significant limitations.

[0003] Current solutions have obvious shortcomings in data processing. Many methods only focus on a single type of data, such as consumption records or health indicators, ignoring the diversity and dynamics of mountaineer behavior. For example, some systems only record snack purchase quantities, but cannot combine real-time physical conditions or social interaction patterns of mountaineers, resulting in a lack of targeted supply recommendations. In addition, the heterogeneity of data sources makes integrated analysis difficult, and the formats of indicators collected by different devices are not uniform, making it difficult to form a unified analysis framework. This fragmented data processing approach limits the system's comprehensive understanding of mountaineer needs.

[0004] The deeper challenge lies in the time sequence misalignment problem. In high-altitude environments, mountaineers' snack consumption behavior, physical condition, and social interaction change rapidly over time. For example, a mountaineer may need high-calorie snacks in the morning due to high-intensity climbing, but in the afternoon due to fatigue, the demand shifts to low-sugar supplements. However, existing systems often fail to accurately capture these changes in demand at these time points, let alone correlate these changes with mountaineers' social behavior, such as the habit of sharing snacks with teammates, which may affect individual supply quantities. This time sequence misalignment phenomenon makes data analysis unable to accurately reflect the actual needs of mountaineers.

[0005] Therefore, how to integrate multi-source heterogeneous data and calibrate time sequence misalignment in a dynamically changing high-altitude environment to accurately capture mountaineers' snack consumption needs has become a key problem in health management and resource allocation. SUMMARY

[0006] The present application provides a health light snack market data analysis method and system based on deep learning, aiming to solve the technical problems of processing efficiency, resource allocation, and time sequence consistency of multi-field data such as snack preference, fitness trajectory, and medical indicators in extreme environments such as high-altitude mountaineering.

[0007] In a first aspect, the application provides a health light snack market data analysis method based on deep learning, which comprises: S101, obtaining multi-field data of mountaineers, including snack consumption frequency, fitness trajectory and medical indicators; S102, clustering the multi-field data to obtain classified data groups; S103, determining high-priority data flow according to the classified data groups; S104, generating synchronized time sequence based on the high-priority data flow; S105, detecting misplacement points in the synchronized time sequence to obtain error-free data chain; S106, fusing snack category preferences and interactive object types in the error-free data chain to generate balanced load indicators; S107, extracting node connections of effect network according to the balanced load indicators to determine multi-level effect structure; S108, simulating data flow control scenarios through the multi-level effect structure to generate distribution priority sequence.

[0008] In a second aspect, the application provides a health light snack market data analysis system based on deep learning, which comprises: A data acquisition module is configured to acquire multi-field data of mountaineers, including snack consumption frequency, fitness trajectory and medical indicators; A data clustering module is configured to cluster the multi-field data to obtain classified data groups; A priority determination module is configured to determine high-priority data flow according to the classified data groups; A time sequence synchronization module is configured to generate synchronized time sequence based on the high-priority data flow; A misplacement detection module is configured to detect misplacement points in the synchronized time sequence to obtain error-free data chain; A load fusion module is configured to fuse snack category preferences and interactive object types in the error-free data chain to generate balanced load indicators; A structure determination module is configured to extract node connections of effect network according to the balanced load indicators to determine multi-level effect structure; A scenario simulation module is configured to simulate data flow control scenarios through the multi-level effect structure to generate distribution priority sequence.

[0009] Compared with the prior art, the technical scheme of the application has at least the following advantages: 1. Through the clustering processing of multi-field data, time sequence synchronization and misplacement detection, the redundancy and errors in the data processing process are effectively reduced, and the efficiency of data analysis is improved, especially suitable for data processing needs in high altitude or extreme environment.

[0010] 2. Through load fusion and balanced index calculation, intelligent scheduling of data flow is realized. When the data resource occupation exceeds the threshold, low priority data is transferred to the standby buffer in time, ensuring the stability and real-time performance of high priority data flow, and optimizing resource allocation.

[0011] 3. Through misplacement detection and dynamic adjustment mechanism, the consistency of time sequence data is ensured, and the influence of potential misplacement points is reduced, thereby improving the accuracy and reliability of data.

[0012] 4. When the behavior log length and social interaction mode are fused in the effect network, the behavior path analysis of mountaineers in snack consumption is strengthened, thereby improving the stability of the effect network and enhancing the adaptability of the model to the actual scene.

[0013] 5. Through the simulation of multi-level effect structure and the optimization of data flow control scene, the resource allocation under different data flow control scenes can be effectively simulated, and the optimized allocation priority sequence is generated according to the dynamic adjustment period, thereby improving the stability and scalability of the system. BRIEF DESCRIPTION OF DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can obtain other drawings based on these drawings without creative labor.

[0015] Figure 1 The flowchart of the health light snack market data analysis method based on deep learning of the present application; Figure 2 The schematic diagram of the health light snack market data analysis system based on deep learning of the present application. DETAILED DESCRIPTION

[0016] The terms "first", "second", "third", "fourth" and the like in the description and in the claims of the present application and above-mentioned drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the use of the terms so construed herein can be interchanged, under suitable circumstances, without changing the meaning of the description by the embodiments described herein. Moreover, the terms "comprising" or "having" and any variations thereof, are intended to cover a non-exclusive inclusion, for example, a process, method, system, product or apparatus that comprises a list of steps or units not necessarily limited to those explicitly stated, but can include other not expressly listed or inherent to such processes, methods, products or apparatus.

[0017] For the sake of understanding, the specific flow of the embodiments of the present application is described below, Figure 1 The flow chart of one embodiment of the health snack market data analysis method based on deep learning provided by the present application is shown, which specifically includes the following steps: S101, obtain multi-field data of mountaineers, including snack consumption frequency, fitness trajectory and medical indicators.

[0018] In a specific embodiment, the process of performing step S101 can specifically include the following steps: Collect the snack consumption frequency statistics, fitness trajectory and medical indicators of high-altitude mountaineers from multi-field sources, wherein the snack consumption frequency statistics are used to identify the snack supplement mode during the mountaineering process; Obtain the original data stream from multi-field sources through a pre-established data interface; Preprocess the original data stream to obtain standardized multi-field data; According to the standardized multi-field data, extract a feature vector including a consumption frequency feature, a trajectory feature and a medical feature; Clean the feature vector, eliminate outliers, generate input data for clustering processing, and store the input data to a pre-set data buffer.

[0019] Specifically, snack consumption frequency, fitness trajectory, and medical indicator data of high-altitude mountaineers are collected through multi-domain data sources. In this step, snack consumption frequency statistics are used to identify snack replenishment patterns during the mountaineering process, which is crucial for determining the energy consumption and replenishment behavior of mountaineers. For example, during the mountaineering process, the energy demand of mountaineers in high-altitude environments is closely related to the frequency of snack replenishment, and changes in snack consumption frequency can help predict the health status and supply needs of mountaineers, especially the supply patterns related to altitude changes. At the same time, fitness trajectory and medical indicators reflect the physical exertion and health status of mountaineers, including physiological indicators such as heart rate and blood oxygen saturation, which help further analyze health risks in high-altitude environments, such as hypoxia and dehydration.

[0020] After data collection, raw data streams are seamlessly obtained from various data sources (such as wearable devices, satellite positioning systems, and medical monitoring instruments) through pre-established data interfaces, ensuring real-time and comprehensive data. Raw data streams are transmitted in real time to the system and standardized before entering the next step of processing. Standardization processing unifies data formats and eliminates dimensional differences, allowing data from different sources to be standardized into a standard format, thereby improving the accuracy and efficiency of subsequent processing. The standardization techniques in this step, especially when dealing with snack consumption frequency, fitness trajectory, and medical indicators, ensure consistency among different dimensions of data, laying the foundation for the next step of feature extraction.

[0021] After data standardization, feature vectors are extracted based on standardized multi-domain data. Feature vectors include consumption frequency features, trajectory features, and medical features, comprehensively reflecting the snack consumption behavior, physical exertion patterns, and health status of mountaineers. Consumption frequency features can include average daily snack intake frequency, snack replenishment frequency during a specific time period, etc.; trajectory features include total travel distance, altitude change, and climbing speed; and medical features reflect physiological data such as heart rate and blood oxygen saturation. The extraction of these feature vectors provides high-quality input data for subsequent data analysis and clustering processing, and is the key to analyzing the relationship between snack replenishment patterns and health status. For example, consider a mountaineer's data: consumption frequency features are 5 times per day, trajectory features are 15 kilometers of daily travel distance, and medical features are an average heart rate of 120 beats per minute. Through extraction, these features form the vector [5, 15, 120], which is used to identify similar mountaineer clusters, enhancing the practical value of data in health snack recommendations.

[0022] Next, the feature vectors are cleaned of outliers. By using statistical methods such as Z-score method, it can effectively identify and remove the error data caused by equipment failure, data loss or other abnormal reasons. This cleaning step ensures the quality of the input data for clustering analysis, avoids the adverse effects of abnormal data on the analysis results, and improves the accuracy and reliability of the analysis. The input data after data cleaning is stored in the preset data buffer. The data buffer ensures the continuity and efficient access of data, so that subsequent clustering processing and analysis can be quickly performed without the need to retrieve data again. In this process, the storage of data is sorted according to the timestamp, ensuring the consistency of data in time, facilitating the timing and logic of subsequent analysis.

[0023] The technical solution solves the timeliness, accuracy and consistency problems that may be encountered when collecting and processing multi-source data in extreme environments, ensuring the reliability and effectiveness of data analysis, and further providing more accurate decision support for health snack market analysis. These technical features together constitute a complete multi-field data collection, processing and analysis framework, providing an innovative technical solution for personalized snack supplement recommendation and efficient market data analysis.

[0024] S102, clustering processing is performed on the multi-field data to obtain classified data groups.

[0025] In a specific embodiment, the process of step S102 can specifically include the following steps: heterogeneous classification of input data using clustering algorithms to generate multiple data groups; respectively calculating the mean and variance of the consumption frequency of each data group, and determining the clustering center of each data group according to the mean and variance of the consumption frequency; grouping multi-field data based on the clustering center to obtain classified data groups; verify the separation degree of the classified data groups, and determine whether the classified data groups meet the preset separation degree threshold, if yes, output the classified data groups.

[0026] Specifically, by collecting high-altitude mountaineer health snack consumption frequency statistics, fitness trajectories, and medical indicators from multi-domain sources, feature vectors are extracted, including consumption frequency values, trajectory coordinates, and medical indicators such as heart rate values. A K-means clustering algorithm, a distance-based clustering method that assigns data points to the nearest cluster center to form groups, is used to classify these feature vectors by heterogeneity. The heterogeneity classification takes into account the dimensional differences of data from different fields, such as consumption frequency being numerical, fitness trajectory being spatial, and medical indicators being physiological. By standardizing them to make them comparable, multiple data groups are generated, each representing a subset of mountaineers with similar snack supplement patterns.

[0027] After each data group is generated, the mean and variance of the consumption frequency of each group are calculated. The key role of this step is to quantify the pattern and variation of snack consumption within each group. For example, in a particular group, if the consumption frequency is more concentrated, the mean is more accurate and the variance is smaller; if the consumption frequency varies greatly, the variance increases accordingly. These statistics can reflect the concentration and dispersion of data within the group, which helps determine the location of the cluster center in the subsequent steps. Based on the mean and variance of the consumption frequency, the cluster center of each data group is further calculated. Exemplarily, center = mean + 0.5 x variance. This way of determining the center can reflect the stability of the consumption pattern and help identify reliable snack supplement patterns in high-altitude environments. The calculation of the cluster center is not only based on the mean, but also takes into account the variance: for groups with smaller variance, the cluster center is closer to the mean, while for groups with larger variance, the cluster center deviates from the mean by a moderate amount of standard deviation adjustment. This approach can more accurately reflect the stability and dependence of snack consumption patterns within the group.

[0028] Based on the determined cluster centers, the algorithm reassigns data points to the nearest cluster center, forming new data groups. This process is the core part of the clustering algorithm, which ensures that multi-domain data is effectively grouped based on similarity, so that the data features within each group are as consistent as possible, and the differences between groups are more obvious. To ensure the effectiveness of clustering, the separation degree of the classified data groups, i.e., the difference between groups, also needs to be verified. In this process, the separation degree is usually evaluated by calculating the ratio of the distance between groups and the distance within groups. If the distance between groups is greater than the distance within groups and greater than the preset separation threshold, it indicates that the clustering result is effective and can distinguish different snack consumption patterns and mountaineer health needs.

[0029] If the preset separation threshold is met, the system will output the classified data groups. These data groups represent the snacking patterns of high-altitude mountaineers in different environments and the characteristics related to their health status. The output results not only help identify and optimize snacking strategies, but also provide a basis for subsequent health management and personalized recommendation systems. For example, one group may represent a high dependence on high-calorie snacks in high-altitude environments, while another group may indicate a balanced snacking pattern in low-altitude environments.

[0030] The clustering processing technology effectively solves the data processing problem in high-altitude mountaineer health management. By standardizing and clustering different types of data, it not only improves data processing efficiency, but also makes health analysis more accurate and targeted, ensuring that the snack recommendation system can be optimized according to specific health status and environmental conditions, providing more suitable health snack recommendations for mountaineers, and ultimately achieving intelligent health management.

[0031] S103, determining a high-priority data stream according to the classified data groups.

[0032] In a specific embodiment, the process of performing step S103 can specifically include the following steps: Calculate the update frequency and data volume size of each classified data group respectively; Based on the update frequency, use a priority queue to sort the classified data groups and generate an initial priority sequence; According to the initial priority sequence, extract the target data group whose update frequency is higher than the preset frequency threshold; Fuse the snack category preferences in the target data group to determine the dependence degree of mountaineers on energy snacks in the target data group; Adjust the initial priority sequence according to the dependence degree to generate a high-priority data stream.

[0033] Specifically, by calculating the update frequency and data volume size of the classified data groups, the processing efficiency of the data stream is further optimized. First, the number of updates of each data group is counted within a set time window, and the number of updates is divided by the length of the time window to obtain the update frequency of each group. At the same time, the data volume size is calculated by accumulating the number of bytes of data in each group. Update frequency and data volume size are important indicators reflecting the activity and data size of the group, which helps determine the priority of the data group in processing.

[0034] The priority queue is maintained by a minimum heap structure, and the update frequency is inserted into the queue as the key value. The initial priority sequence is outputted by sorting the frequencies from high to low, ensuring that the groups with higher update frequencies are placed at the front of the queue. During the sorting process, groups with higher frequencies are processed first, improving the system's response to real-time data streams. This sorting process generates the initial priority sequence.

[0035] After obtaining the initial priority sequence, the target data groups are further filtered based on a preset frequency threshold. Target data groups refer to groups with update frequencies exceeding the threshold. For example, when the preset threshold is 5 times per hour, groups with update frequencies higher than this threshold are considered target data groups. Through this filtering process, groups with higher activity levels can be quickly identified, allowing resources to be concentrated for priority processing.

[0036] Further, after determining the target data groups, the zero food class preference data in the group is fused to analyze the dependence of mountaineers on different zero food classes. Zero food class preference refers to the consumption tendency of mountaineers towards specific snack types (such as energy bars, nuts, etc.) in high-altitude environments. This preference data is obtained through snack consumption frequency statistics. For example, mountaineers may have a higher preference for energy snacks than other types of snacks. Based on this preference, a weighted summation method is used to calculate the dependence of target data groups on energy snacks. The higher the dependence, the stronger the group's demand for energy supplementation in high-altitude environments. For example, the consumption frequency of energy snacks is multiplied by a preset weight of 0.7, and other types of snacks are multiplied by 0.3, and the sum is the dependence score. The higher the score, the greater the dependence on energy snacks. This calculation takes into account the urgency of energy supplementation in high-altitude mountaineering, improving the relevance of data processing.

[0037] To further improve the accuracy and relevance of data processing, the weights of zero food class preferences are adjusted according to actual needs in different mountaineering scenarios. For example, in long-distance mountaineering scenarios, the weight of energy snacks can be increased to 0.8 to better reflect mountaineers' demand for rapid energy recovery. Through this adjustment, mountaineers' snack consumption patterns can be more accurately identified and optimized, ensuring that data processing can respond to mountaineers' energy supplementation needs in real time.

[0038] Finally, the initial priority sequence is adjusted according to the dependency degree to generate a high-priority data stream. By multiplying the dependency degree as an adjustment factor with the frequency value in the initial priority sequence, the adjusted priority value is obtained. According to the adjusted priority value, the high-priority data stream is reordered. This adjustment ensures that groups with high dependency degree are given priority in data stream processing, especially when the mountaineer's medical indicators show potential risks such as hypoglycemia or energy deficiency, the group with high dependency degree will be given priority, thus providing the required snack supplement information in time.

[0039] The application of this technical solution in high-altitude mountain climbing data processing can effectively improve the processing efficiency of data stream and avoid health risks caused by delayed processing. By combining factors such as update frequency, data volume, snack preference and dependency degree, the system can ensure real-time response to the health and energy needs of mountaineers in high-altitude environments. This significantly improves the accuracy and real-time nature of snack consumption pattern analysis, providing personalized and healthy snack recommendations for mountaineers, optimizing resource allocation and improving the effectiveness of health management.

[0040] S104, generating a synchronized time sequence based on the high-priority data stream.

[0041] In a specific embodiment, the process of step S104 can specifically include the following steps: Obtain the timestamp information of the high-priority data stream; If the timestamp information shows real-time update characteristics, immediately allocate computing resources to process the high-priority data stream; Align the high-priority data stream in time sequence through computing resources to generate an initial time sequence; Check if the purchase time distribution in the initial time sequence is aligned with the mountain climbing track. If yes, generate a synchronized time sequence. If not, adjust the timestamp of the initial time sequence to generate a synchronized time sequence.

[0042] Specifically, the timestamp information of the high-priority data stream is obtained, which is usually extracted by parsing the metadata field in the data packet header, ensuring that each data point has a time marker accurate to the second. In order to support subsequent real-time judgment, all timestamps are standardized to coordinated universal time format. The acquisition and formatting of timestamps is an important step to ensure the consistency and real-time nature of multi-source data. In high-altitude mountain climbing environments, these timestamps may come from the mountaineer's wearable device, recording the GPS time embedded when recording snack consumption frequency. Such processing can effectively reduce data processing delay and improve system response speed.

[0043] After obtaining the timestamp information, if the real-time update feature is displayed (i.e., the interval between timestamps is less than a preset threshold such as 5 seconds), computing resources are immediately allocated for data stream processing. The allocation of computing resources is accomplished by activating cloud servers or local computing nodes, which may specifically include allocating at least 2 CPU cores and 1 GB of memory to meet the processing needs of real-time data streams. In high-altitude mountaineering activities, such as when mountaineers report snack consumption in real time through an APP, frequent updates of timestamps mean that the system needs to respond quickly to avoid analysis delays caused by data accumulation. Through the rapid allocation and monitoring of resource pools, computing resources are ensured to be in place within 1 second, thereby enabling immediate processing of real-time data streams and ensuring the efficient operation of the system.

[0044] Subsequently, using the allocated computing resources, the high-priority data stream is time-aligned. Time alignment is achieved by filling in the discontinuous parts of the timestamps using linear interpolation, for example, inserting a reasonable average value for the missing 10-second data. Through a sliding window algorithm, the data is smoothed to further eliminate the influence of sudden data points, ensuring the continuity of the time series data. This time smoothing helps ensure the consistency of snack consumption time and mountaineering trajectory data, thereby improving the reliability and accuracy of the data stream, especially when processing multiple group data, which can refine the dependency of each group and provide more accurate information for subsequent analysis and decision-making.

[0045] After generating the initial time series, it is checked whether the purchase time distribution is aligned with the mountaineering trajectory. This process involves calculating the statistical distribution of purchase times and performing correlation analysis with the time points of the mountaineering trajectory data, using the Pearson correlation coefficient method for comparison. When the correlation coefficient is greater than a preset threshold (such as 0.8), it indicates that the purchase time distribution is aligned with the mountaineering trajectory, and the synchronized time series can be generated directly. If the correlation is low, potential misalignment points are marked and the timestamps are adjusted to ensure the synchronization of the time series.

[0046] The timestamp adjustment process identifies the misalignment offset and uses the least squares method to fit the offset curve to correct each timestamp. For example, the average deviation value (such as 2 seconds) can be calculated and adjusted to ensure that the alignment degree of the corrected time series reaches the preset threshold. If the corrected sequence is successfully aligned, the synchronized time series is output. In addition, during the correction process, other data sources such as social sharing times can be integrated to further enhance the consistency and accuracy of the time series data. Through this adjustment process, the system can optimize the snack replenishment mode according to the real-time activities and needs of mountaineers, ensuring the accuracy and real-time nature of the data stream.

[0047] The technical solution significantly improves the analysis efficiency and real-time performance of healthy snack consumption patterns in high-altitude mountaineering scenarios. Through precise time alignment and timestamp correction, the energy supplement needs of mountaineers can be reflected in real time, and the health risks caused by data deviation can be effectively avoided. This solution helps to improve the response speed of the system and the accuracy of resource allocation, optimizes the recommendation and replenishment mode of snack food during mountaineering, and ensures the efficiency and accuracy of data processing.

[0048] S105, detecting the dislocation point in the synchronized time sequence to obtain a dislocation-free data chain.

[0049] In a specific embodiment, the process of performing step S105 can specifically include the following steps: Analyze the synchronized time sequence by a preset dislocation detection algorithm to determine potential dislocation points; For potential dislocation points, a dynamic adjustment mechanism is used to rearrange the batch data in the synchronized time sequence to generate an adjusted time sequence; Fusion of the number of social shares in the adjusted time sequence, calibration of the time sequence consistency of snack experience exchange among mountaineers; Verify whether the time sequence after calibration of time sequence consistency contains dislocation points, if not, output a dislocation-free data chain.

[0050] Specifically, in the implementation of step S105, the process of detecting dislocation points in the synchronized time sequence to generate a dislocation-free data chain, first analyzes the synchronized time sequence by a preset dislocation detection algorithm to identify potential dislocation points. Specifically, by extracting the timestamp information in the synchronized time sequence and the related consumption frequency statistical features, the system can identify the inconsistency in the time sequence data. The consumption frequency statistical features reflect the regularity of snack consumption in a certain time period, and the timestamp marks the time of data collection, which together provides the complete context of the time sequence data. On this basis, a sliding window method is used to scan the time sequence, and the stability of the sequence is judged by calculating the continuity index of the timestamp in each window. Specifically, when the standard deviation of the timestamp in the sliding window exceeds the preset threshold, the window is marked as a potential dislocation point. This method effectively identifies the data delay or signal instability problem caused by high-altitude environment, avoiding the influence of data dislocation. Illustratively, in the high-altitude mountaineering scenario, assuming that the synchronized time sequence contains 100 data points, the timestamp is from 1 to 100 seconds, and the consumption frequency statistics show that energy snacks are supplemented every 10 seconds. By the sliding window method, set the window size to 5, calculate the standard deviation of each window, if the standard deviation of a certain window is 0.8 which exceeds the threshold 0.5, it is marked as a potential dislocation point. This method can early detect the delay caused by weak high-altitude signals in the mountaineering track, which is beneficial to timely adjust the data processing resources.

[0051] For potential misalignment points, a dynamic adjustment mechanism is further adopted to rearrange the batch data in the synchronized time series to optimize the integrity of the time series. By marking the location of potential misalignment points, the time series is divided into multiple batches, each corresponding to a continuous data segment. Then, the update frequency and volume size of each batch are calculated, the update frequency reflects the average value of the timestamp interval within each batch, and the volume size is the number of data points in the batch. The batches are sorted using a priority queue, with high-priority batches being those with higher update frequency and alignment with the mountaineering trajectory. Through this sorting, high-priority batches can be placed at the front of the sequence, thereby avoiding the delay impact of low-frequency data. This dynamic adjustment mechanism ensures the time series integrity of snack consumption records and can effectively support subsequent analysis, such as energy supplement patterns and health demand analysis of mountaineers. Exemplarily, a mountaineering group's data, the potential misalignment points divide the sequence into 3 batches, batch 1 has an update frequency of every 2 seconds and a volume size of 20 points; batch 2 has an update frequency of every 5 seconds and a volume size of 15 points; batch 3 has an update frequency of every 3 seconds and a volume size of 10 points, after sorting using the priority queue, batch 1 is placed at the front, and a new sequence is formed by rearrangement. This adjustment ensures that the snack supplement pattern aligns with the fitness trajectory, avoiding the impact of misalignment on dependency analysis, and is beneficial to optimizing the resource allocation model.

[0052] After the adjusted time series sequence is generated, the social sharing frequency data is fused to calibrate the time series consistency of snack experience exchange among mountaineers. Specifically, the social sharing frequency in the adjusted time series sequence is extracted, which reflects the frequency of mountaineers sharing snack experience on social platforms, and the correlation strength between sharing frequency and snack consumption frequency is calculated by Pearson correlation coefficient. If the correlation strength is higher than a preset threshold (such as 0.7), it means that the snack consumption and social sharing have strong consistency, and the system will adjust the weight of related timestamps to align the sharing events with the consumption events. Through this process, the time series consistency can be effectively calibrated to ensure the coordination of snack consumption data and social exchange data in time series, further enhancing the stability of the data chain.

[0053] Preferably, when fusing social sharing frequency, the behavior log duration can also be considered, which refers to the duration of recording mountaineering social interaction, and is weightedly averaged with the sharing frequency, exemplarily, the sharing frequency weight is 0.6 and the log duration weight is 0.4. Subsequently, the time series consistency is further calibrated to improve the accuracy in the multi-level effect structure.

[0054] After generating the adjusted time series and calibrating its consistency, the adjusted time series is verified again to check for any remaining misalignments. The continuity metrics of the sequence are rechecked using a sliding window method. If all window metrics are above a preset threshold, the adjusted time series is confirmed to be free of misalignments. Finally, if the adjusted time series passes verification, a misalignment-free data chain is output as the basis for subsequent high-priority data stream processing. This process ensures the high quality and accuracy of the entire data chain, avoiding data processing errors caused by misalignments, thus providing reliable data support for in-depth analysis of the healthy light snack market.

[0055] This technical solution significantly improves the accuracy and real-time performance of data processing, especially in high-altitude environments, effectively resolving misalignment issues caused by signal delays or inconsistent data acquisition. Through misalignment detection and dynamic adjustment mechanisms, it ensures the stability and consistency of time-series data, providing precise data support for further analysis of healthy snack consumption patterns. The beneficial effect of this method lies in improving the synchronous processing capability of multidimensional data (such as snack consumption and social sharing), thereby optimizing mountaineers' snack replenishment patterns and health management.

[0056] S106. Integrate snack category preferences and interaction object types from the error-free data chain to generate a balanced load index.

[0057] In one specific embodiment, the process of executing step S106 may specifically include the following steps: Retrieve snack category preferences and interaction object types from a seamless data chain; The preset fusion algorithm is used to weight and combine snack category preferences and interaction object types to generate fused data; If the resource consumption of the fused data exceeds the preset resource consumption threshold, then the low-priority data is moved to the backup buffer. Extract fused data that exceeds the preset resource usage threshold and calculate the load balancing index accordingly.

[0058] Specifically, the study extracts climbers' snack preferences and interaction types in high-altitude environments from a misaligned data chain. Snack preferences indicate the climbers' dependence on energy-rich snacks, while interaction types refer to the types of people they interact with on social media platforms, such as teammates or online community members. This information is derived through time-series data analysis, ensuring that after effectively removing misaligned points, the remaining data chain provides accurate records of snack consumption patterns and social behaviors.

[0059] The preset fusion algorithm is used to weight and combine the zero food preference and the interactive object type. Through weighted summation, the weight values of the zero food preference and the interactive object type are multiplied, and the comprehensive fusion data is obtained. For example, the zero food preference can give a higher weight to energy bars based on the consumption frequency, while the weight of snacks such as nuts is lower. The interactive object type is weighted according to the importance of social interaction, for example, interaction with teammates may give a higher weight, while interaction with online communities may give a lower weight. Such weighted combination can accurately reflect the behavior patterns of mountaineers in different social and consumption scenarios, ensuring the correct docking and fusion of various data in the time sequence chain.

[0060] After obtaining the fused data, it is necessary to detect its resource occupation and confirm whether the data exceeds the preset resource occupation threshold. If the resource occupation is too high, it may cause system processing delay or resource bottleneck. Therefore, based on the resource occupation of the fused data, if it is found that it exceeds the threshold, the system will trigger a dynamic adjustment mechanism to transfer low-priority data (such as data that does not greatly affect the analysis of snack replenishment patterns) to a backup buffer. This processing not only optimizes the processing efficiency of the data stream, but also effectively avoids data processing failure or delay caused by excessive resource occupation, ensuring smooth processing of high-priority data.

[0061] By transferring low-priority data and allocating resources, it can ensure that the remaining data is processed efficiently according to priority, avoiding interference of complex data streams in high-altitude environments. Thereafter, the balanced load index is calculated based on the fused data. The calculation of the balanced load index is based on the update frequency and volume size of the data, reflecting the resource consumption of the data stream in the system. Specifically, if the occupation rate of the fused data is high, the balanced load index value will be large, indicating that the system resource load is heavy, and the data stream needs to be adjusted to maintain stable operation of the system. Exemplarily, balanced load index = (total amount of fused data / resource capacity) x 100, assuming that the update frequency of a data group is once per hour, the volume of fused data is 500MB, and the resource capacity is 1GB, then the balanced load index is calculated as 48.8%.

[0062] For example, in a high-altitude mountaineering scenario, if the data indicates that the energy snack dependency is 0.85 and the interactive object type is teammates (with a weight of 1.0), the generated balanced load index can reflect the current data load status of the system. If the index value is too high, the system can optimize resource allocation through priority adjustment or buffer data transfer. Ultimately, the balanced load index calculated from these calculations will be stored in the preset index database for subsequent analysis and decision-making, such as adjusting the resource allocation model or improving the dynamic adjustment mechanism.

[0063] The technical solution can refine the understanding of the snacking consumption pattern and social behavior of mountaineers in high-altitude environments by weighted combination of zero food preference and interactive object type, and further accurately analyze the energy supplement demand and social interaction path. At the same time, through the calculation and resource allocation of the balanced load index, efficient data flow processing is ensured, and system overload caused by high occupancy data is avoided, and the stability and accuracy of data analysis are optimized. These technical means jointly solve the resource consumption problem in high-altitude mountaineering data flow processing, improve the data analysis efficiency, reduce the delay, and maintain the efficient stability of the data flow in the multi-level effect structure.

[0064] S107, extracting node connection of effect network according to balanced load index, determining multi-level effect structure.

[0065] In a specific embodiment, the process of performing step S107 can specifically include the following steps: extracting node connection information from the balanced load index; using a clustering algorithm to analyze the association strength of the node connection information, generating an association strength matrix between nodes; determining the multi-level effect structure of the effect network according to the association strength matrix; fusing the behavior log duration in the multi-level effect structure, strengthening the influence path of the mountaineering social mode on snack consumption, and obtaining the strengthened multi-level effect structure; verify the stability of the strengthened multi-level effect structure, if it meets the preset stability threshold, output the strengthened multi-level effect structure.

[0066] Specifically, by combining the balanced load index and the relationship of the effect network nodes, a multi-level effect structure is generated to analyze the interaction mode between snack consumption and behavior logs of high-altitude mountaineers. The process first extracts node connection information from the balanced load index, which reflects the relationship between the snack consumption patterns of each mountaineer. These information is extracted by analyzing parameters such as resource occupancy rate and data flow priority. By analyzing these data distributions, potential relationships between snack consumption behavior and interaction mode among mountaineers can be identified. The node connection information includes the frequency of snack consumption, the type of social sharing objects, and the consumption patterns of mountaineers in different environments.

[0067] The clustering algorithm is used to analyze the connection information of these nodes and generate a correlation strength matrix between nodes. The K-means clustering algorithm is usually used to process the node connection information, calculate the Euclidean distance between nodes, and update the clustering center through multiple iterations to finally obtain the correlation strength between nodes. The correlation strength of each node reflects the behavior and interaction of different hikers in snack consumption. For example, if the consumption frequency of some nodes is highly consistent, their correlation strength is higher, indicating that these nodes have strong similarity in snack consumption patterns, which may be due to the same hiking team or similar health needs.

[0068] An effect network refers to the network structure of the interaction and influence relationship between different elements or nodes in a complex system. In an effect network, each node represents an entity or event, and the connection between nodes represents their mutual influence or dependence. Effect networks can reveal how these nodes collectively produce certain effects or results through their interactions, especially in a multi-level and dynamic environment.

[0069] After generating the correlation strength matrix, a multi-level structure of the effect network is constructed based on the matrix. The hierarchical division of the effect network is based on the strong connection edges in the correlation strength matrix. Further application of community detection algorithms (such as the Louvain algorithm) can identify the multi-level structure in the effect network by optimizing the modularity. In this structure, the bottom level corresponds to the basic snack consumption pattern, the middle level integrates the information of the fitness track, and the upper level reflects the influence of external factors such as medical indicators on the snack consumption pattern. In this way, the nodes in the network are divided into multiple levels, forming a hierarchical structure that better demonstrates the influence path of different factors on the snack consumption pattern.

[0070] To strengthen the influence path of the hiking social mode on snack consumption, the behavior log duration is further integrated as a weight factor into the multi-level effect structure. The behavior log duration reflects the activity intensity and interaction frequency of each node, and nodes with higher duration have stronger influence in the network. Therefore, by adjusting the effect path in the network with duration as a weight factor, the influence of social interaction on snack consumption patterns can be further highlighted, ensuring the timing consistency of social sharing. For example, if the social interaction duration of a node is longer, the strength of its influence path will be increased, further enhancing its role in the network, thereby improving the understanding and prediction accuracy of snack consumption patterns in high-altitude environments.

[0071] To ensure the effectiveness and stability of the structure, it is necessary to verify the stability of the multi-level effect structure. This verification process is evaluated by calculating the variance of the effect path. If the variance of some paths in the network is lower than the preset stability threshold, the effect network is considered stable and can accurately reflect the snack consumption pattern of high-altitude mountaineers. After verification, the structure will be output and used for subsequent resource allocation model optimization, ensuring the efficiency and accuracy of the processing of mountaineer snack consumption records.

[0072] Through this series of steps, the snack consumption pattern of mountaineers in high-altitude environments and social behavior can be effectively combined to build a hierarchical and dynamically adjusted effect network. This network not only accurately reflects the dynamic changes of snack consumption, but also adjusts its structure according to different social patterns, consumption frequency, and behavior log length, thereby improving the accuracy of overall data processing, optimizing resource allocation, and reducing the risk of delay in high-altitude environments.

[0073] S108, simulate the data flow control scenario through the multi-level effect structure to generate an allocation priority sequence.

[0074] In a specific embodiment, the process of executing step S108 can specifically include the following steps: Obtain the transition path of snack preference to interactive object type in the multi-level effect structure, and build a data flow control model through the transition path; Use the data flow control model to simulate the data flow control scenario and generate an initial allocation priority sequence; Optimize the resource allocation model according to the initial allocation priority sequence and the dynamic adjustment period to generate an optimized allocation priority sequence.

[0075] Specifically, by analyzing the transition path of snack preference to interactive object type in the multi-level effect structure, key transition path data is obtained. The transition path is composed of multiple nodes and their associations, such as snack preference nodes, social sharing nodes, and interactive object type nodes. By analyzing the association strength between snack preference nodes and interactive object type nodes through clustering methods, key paths that affect data flow control can be extracted, and the weight value of the path can be defined.

[0076] Based on the obtained transition paths, a data flow control model is constructed. The core of this model is a graph-based modeling method, where nodes represent data elements (such as snack food preferences, interactive object types, etc.), and edges represent the flow relationship between these nodes. Specifically, each node of the transition path is mapped to a node in the data flow, and the edges between these nodes correspond to the relationship of data flow. The data flow control model uses a weighted graph to represent, where the weight is determined by the associated strength of the path, such as the strength value between nodes calculated by the K-means clustering method. The model further defines the priority rules for data flowing from the snack food preference node to the interactive object type node, and calculates the flow efficiency of each data flow using the formula of path length.

[0077] The data flow control model is used to simulate the data flow control scenario and generate an initial allocation priority sequence. This step ensures that the allocation of each snack category is sorted according to its priority by comparing the simulation scenario with the actual data flow. For example, by inputting the time series data of snack food preferences and interactive object types, the flow in different groups is simulated. During the simulation process, the real-time timestamp information ensures that high-priority snack food categories can be processed in a timely manner, thereby avoiding data misplacement and ensuring the time consistency of the model.

[0078] After generating the initial allocation priority sequence, the resource allocation model is further optimized based on a dynamic adjustment period. The key to this optimization step is to dynamically adjust the resource allocation of each group based on the priority of the initial sequence, combined with social behavior log data and actual demand for snack consumption. For example, when the model identifies that some groups have higher demand for snack consumption in high-altitude environments, it adjusts the priority of these groups to ensure that they can obtain the required resources in priority. Through this dynamic adjustment mechanism, an optimized allocation priority sequence is generated. This optimized sequence not only considers the number of social shares and the length of behavior logs, but also combines the time consistency of mountaineer snack consumption behavior, effectively improving the rationality and real-time performance of resource allocation.

[0079] After these steps, the optimized allocation priority sequence is stored in a preset priority database for subsequent access and processing. In high-altitude mountain environments, this optimized allocation priority sequence can effectively avoid data delay and misplacement, thereby improving the processing efficiency of high-altitude mountaineer health snack consumption records and ensuring the rational allocation of resources. By integrating the time series data of snack food preferences, social sharing behavior, and high-altitude mountain trajectories, a stable and efficient data flow control model is finally formed, providing technical support for the precise allocation of the health snack market.

[0080] The technical means has remarkable technical effects in solving the problems of data flow dislocation, delay and unreasonable resource allocation in the prior art. By combining dynamic adjustment period and multi-level effect structure, the allocation sequence is effectively optimized, and the response speed and accuracy of resource allocation are improved, which is especially suitable for zero food consumption optimization in special environments such as high-altitude mountaineering. In addition, the data flow control model strengthens the influence of social mode while ensuring the synchronization and timing consistency of different data flows, thereby optimizing the stability and running efficiency of the overall system.

[0081] The health light snack market data analysis method based on deep learning in the embodiments of the present application is described above, and the health light snack market data analysis system based on deep learning in the embodiments of the present application is described below. Please refer to Figure 2 An embodiment of the structure schematic diagram of the health light snack market data analysis system based on deep learning provided by the present application is provided, and the system comprises: The data acquisition module 10 is used for acquiring the multi-field data of mountaineers, and the multi-field data includes snack consumption frequency, fitness trajectory and medical indicators.

[0082] The data clustering module 20 is used for clustering the multi-field data to obtain classified data groups.

[0083] The priority determination module 30 is used for determining high-priority data flow according to the classified data groups.

[0084] The timing synchronization module 40 is used for generating a synchronized timing sequence according to the high-priority data flow.

[0085] The dislocation detection module 50 is used for detecting dislocation points in the synchronized timing sequence to obtain a dislocation-free data chain.

[0086] The load fusion module 60 is used for fusing the zero food preference and interactive object type in the dislocation-free data chain to generate a balanced load index.

[0087] The structure determination module 70 is used for extracting node connections of the effect network according to the balanced load index to determine a multi-level effect structure.

[0088] The scene simulation module 80 is used for simulating a data flow control scene through the multi-level effect structure to generate an allocation priority sequence.

[0089] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A deep learning-based health snack market data analysis method, characterized in that, The method comprises: S101, obtaining multi-field data of mountaineers, the multi-field data comprising snack consumption frequency, fitness trajectory and medical indicators; S102, performing clustering processing on the multi-field data to obtain classified data groups; S103, determining high-priority data streams according to the classified data groups; S104, generating synchronized time series based on the high-priority data streams; S105, detecting misplacement points in the synchronized time series to obtain error-free data chains; S106, fusing snack category preferences and interactive object types in the error-free data chains to generate balanced load indicators; S107, extracting node connections of an effect network according to the balanced load indicators to determine a multi-level effect structure; S108, simulating data flow control scenarios through the multi-level effect structure to generate distribution priority sequences.

2. The method of claim 1, wherein, The S101 comprises: collecting snack consumption frequency statistics, fitness trajectories and medical indicators of high-altitude mountaineers from multi-field sources, wherein the snack consumption frequency statistics are used to identify snack replenishment patterns during mountaineering; obtaining original data streams from the multi-field sources through pre-established data interfaces; preprocessing the original data streams to obtain standardized multi-field data; extracting feature vectors comprising consumption frequency features, trajectory features and medical features according to the standardized multi-field data; performing data cleaning on the feature vectors to eliminate outliers and generating input data for clustering processing, and storing the input data to a pre-set data buffer.

3. The method of claim 2, wherein, The S102 comprises: performing heterogeneity classification on the input data using a clustering algorithm to generate multiple data groups; calculating the mean and variance of the consumption frequency of each data group, and determining the clustering center of each data group according to the mean and the variance; grouping the multi-field data based on the clustering center to obtain classified data groups; verifying the separation degree of the classified data groups, and determining whether the classified data groups meet a pre-set separation degree threshold, and if so, outputting the classified data groups.

4. The method of claim 1, wherein, The S103 comprises: calculating the update frequency and data volume size of each classified data group; sorting the classified data groups using a priority queue based on the update frequency to generate an initial priority sequence; extracting target data groups with an update frequency higher than a pre-set frequency threshold according to the initial priority sequence; fusing snack category preferences in the target data groups to determine the dependence degree of mountaineers on energy snacks in the target data groups; adjusting the initial priority sequence according to the dependence degree to generate high-priority data streams.

5. The method of claim 1, wherein, The S104 comprises: obtaining timestamp information of the high-priority data streams; if the timestamp information shows real-time update features, immediately allocating computing resources to process the high-priority data streams; performing time series alignment on the high-priority data streams through the computing resources to generate an initial time series; checking whether the purchase time distribution in the initial time sequence is aligned with the mountaineering track, and if so, generating a synchronized time sequence, and if not, adjusting the time stamp of the initial time sequence to generate a synchronized time sequence.

6. The method of claim 1, wherein, The S105 includes: analyzing the synchronized time sequence by a preset misalignment detection algorithm to determine potential misalignment points; for the potential misalignment points, reordering the batch data in the synchronized time sequence by a dynamic adjustment mechanism to generate an adjusted time sequence; fusing the number of social shares in the adjusted time sequence to calibrate the timing consistency of snack experience exchange among mountaineers; verifying whether the timing consistency calibrated time sequence contains misalignment points, and if not, outputting a misalignment-free data chain.

7. The method of claim 1, wherein, The S106 includes: obtaining snack category preferences and interactive object types in the misalignment-free data chain; combining the snack category preferences and the interactive object types by a preset fusion algorithm to generate fusion data; detecting whether the resource occupation of the fusion data exceeds a preset resource occupation threshold, and if so, transferring low-priority data to a backup buffer; extracting fusion data that exceeds the preset resource occupation threshold and calculating an equilibrium load index accordingly.

8. The method of claim 1, wherein, The S107 includes: extracting node connection information from the equilibrium load index; analyzing the node connection information by a clustering algorithm to generate a correlation strength matrix between nodes; determining a multi-level effect structure of an effect network according to the correlation strength matrix; fusing the behavior log duration in the multi-level effect structure to strengthen the influence path of the mountaineering social mode on snack consumption, and obtaining a strengthened multi-level effect structure; verifying the stability of the strengthened multi-level effect structure, and if it meets a preset stability threshold, outputting the strengthened multi-level effect structure.

9. The method of claim 1, wherein, The S108 includes: obtaining a transition path from snack category preferences to interactive object types in the multi-level effect structure, and constructing a data flow control model through the transition path; simulating a data flow control scenario by the data flow control model to generate an initial allocation priority sequence; optimizing a resource allocation model according to the initial allocation priority sequence and a dynamic adjustment period to generate an optimized allocation priority sequence.

10. A deep learning based healthy snack market data analysis system for implementing the method according to any one of claims 1 to 9, characterized in that, The system includes: a data acquisition module for acquiring mountaineer multi-field data, the multi-field data including snack consumption frequency, fitness track and medical indicators; a data clustering module for clustering the multi-field data to obtain classified data groups; a priority determination module for determining high-priority data flow according to the classified data groups; a timing synchronization module for generating a synchronized time sequence according to the high-priority data flow; a misalignment detection module for detecting misalignment points in the synchronized time sequence to obtain a misalignment-free data chain; a load fusion module for fusing snack category preferences and interactive object types in the misalignment-free data chain to generate an equilibrium load index; a structure determination module for extracting node connections of an effect network according to the equilibrium load index to determine a multi-level effect structure; a scenario simulation module for simulating a data flow control scenario through the multi-level effect structure, and generating an allocation priority sequence.

Images