Data temperature evaluation method, device and equipment for heavy haul railway electric service data and medium

Abstract

The application relates to the technical field of railway traffic, in particular to a data temperature evaluation method and device for heavy-load railway electric service data, computer equipment and a storage medium. The method comprises the following steps: obtaining initial hotness evaluation values of a plurality of business data objects in heavy-load railway electric service and correlation relationship parameters between any two business data objects; determining a current hotness value of a target business data object based on access information, a historical hotness value and a time decay model of the target business data object; determining hotness conduction increments of other business data objects associated with the target business data object based on the correlation relationship parameters and the current hotness value of the target business data object, and updating the current hotness values of the other business data objects by combining the historical hotness values of the other business data objects and the time decay model; and performing hierarchical storage management on the plurality of business data objects based on the current hotness values of the business data objects.

Description

Technical Field

[0001] This application relates to the field of railway transportation technology, and more specifically, to a method, apparatus, computer equipment, and storage medium for evaluating the data temperature of heavy-haul railway electrical data. Background Technology

[0002] In the storage and management of massive amounts of heavy-haul railway electrical data, a tiered storage strategy of hot and cold data is typically adopted to balance storage costs and access performance. Existing data popularity assessment methods mainly rely on statistical indicators such as the access frequency and recent access time of individual data objects, calculating data popularity through a preset time decay function, and then determining the data's storage level. However, in complex business systems, different data objects often have strong coupled business logic relationships or time-series dependencies. Existing solutions only assess the popularity of individual data objects in isolation, ignoring the correlation and transmission effects between data. This leads to a situation where, when a core data is frequently accessed, strongly related data may be judged as cold data and migrated to low-speed storage media due to their own low access indicators. This inconsistent pace of popularity decay of related data can easily cause fragmentation problems during cross-layer retrieval, resulting in broken data read links and increased I / O latency in heavy-haul railway electrical data, making it difficult to meet the requirements of high real-time and high-reliability business scenarios for data integrity and access efficiency. Summary of the Invention

[0003] This application provides a method, apparatus, computer equipment, and storage medium for evaluating the data temperature of heavy-haul railway electrical data.

[0004] The first aspect of this application provides a method for evaluating the data temperature of electrical data for heavy-haul railways, including: Obtain the initial heat assessment values ​​of multiple business data objects in the electrical system of heavy-haul railways, as well as the association parameters between any two of the business data objects; In response to the access of a target business data object, the current popularity value of the target business data object is determined based on the access information, historical popularity value, and time decay model of the target business data object; Based on the association parameters and the current popularity value of the target business data object, determine the popularity transmission increment of other business data objects associated with the target business data object, and update the current popularity value of the other business data objects by combining the historical popularity values ​​of the other business data objects and the time decay model; Based on the current popularity value of each of the business data objects, the multiple business data objects are stored and managed in a hierarchical manner.

[0005] In an optional embodiment of this application, obtaining the association parameter between any two of the business data objects includes: A subjective association matrix is ​​obtained based on prior knowledge, and the subjective association matrix contains the subjective association degree between any two of the business data objects. Based on the historical access statistics of the multiple business data objects, an objective correlation matrix is ​​obtained, which contains the objective correlation degree between any two of the business data objects. The subjective correlation matrix and the objective correlation matrix are normalized and then linearly fused according to preset weights to obtain a fused correlation matrix. The association parameters are determined based on the element values ​​in the fusion association matrix.

[0006] In an optional embodiment of this application, obtaining the subjective association matrix based on prior knowledge includes: Obtain the correlation scores between any two business data objects from multiple evaluation nodes, and construct an initial score matrix; Calculate the coefficient of variation of each element in the initial rating matrix. If the coefficient of variation is greater than a preset convergence threshold, trigger re-rating until the coefficient of variation is less than or equal to the preset convergence threshold to obtain the subjective association matrix. The process of obtaining an objective correlation matrix based on historical access statistics of the multiple business data objects includes: Sort the historical access time series data of any two of the business data objects, calculate the rank correlation coefficient, and construct the objective correlation matrix based on the absolute value of the rank correlation coefficient; The linear fusion according to preset weights yields the fusion correlation matrix:

[0007] in, For the fusion correlation matrix, The normalized subjective association matrix, The normalized objective correlation matrix, The preset subjective weight allocation coefficients.

[0008] In an optional embodiment of this application, updating the current popularity value of the other business data objects includes: For any of the other business data objects, obtain its historical popularity value at the previous evaluation time, its basic observation increment at the current evaluation time, and the popularity transmission increment from the target business data object; Based on the historical popularity value, the basic observation increment, and the popularity transmission increment, determine the current popularity value of the other business data objects at the current evaluation time; The historical popularity value decays exponentially with time intervals, and the popularity transmission increment is positively correlated with the correlation parameter and the access weight of the target business data object.

[0009] In an optional embodiment of this application, the current popularity value of the other business data objects at the current evaluation time satisfies the following calculation model:

[0010] in, For the first Other business data objects at the current evaluation time The current popularity value, For the previous assessment time Historical popularity value The coefficient of heat decay rate, for and The time interval between This is the basic observation increment. For the target business data object The association parameter between the first other business data object and the first other business data object. This is the access weight function value of the first business data object at the current evaluation time.

[0011] In an optional embodiment of this application, the method further includes based on the time interval. Execute branch calculation strategy: when If the heat value is less than or equal to the first preset threshold, skip the calculation of the heat value at the current evaluation time; when When the current popularity value is greater than the first preset threshold and less than the second preset threshold, the current popularity value of the target business data object is calculated first, and then the current popularity value of all other business data objects is calculated based on the calculation model. when When the threshold is greater than or equal to the second preset threshold, only the time decay term of the plurality of business data objects is calculated; according to Update the current popularity value.

[0012] In an optional embodiment of this application, obtaining the initial popularity assessment values ​​of multiple business data objects includes: Based on preset multi-dimensional evaluation rules, anonymous scoring results of multiple evaluation nodes for each of the business data objects are obtained. Calculate the coefficient of variation of the anonymous scoring results. If the coefficient of variation is greater than a preset convergence threshold, then the scoring distribution is fed back and a re-scoring is triggered until the coefficient of variation is less than or equal to the preset convergence threshold. The average value of the converged anonymous scoring results is determined as the initial popularity assessment value for the corresponding business data object.

[0013] In one optional embodiment of this application, the business data object includes heavy-haul railway electrical monitoring data, which includes at least one of track voltage data, switch machine status data, track insulation data, signal operating current data, and environmental status data. The hierarchical storage management of the multiple business data objects includes: Based on the preset temperature scoring range, the multiple business data objects are divided into a hot data layer, a warm data layer, and a cold data layer. The hot data layer data is stored in a first performance storage medium, the warm data layer data is stored in a second performance storage medium, and the cold data layer data is stored in a third performance storage medium, wherein the read / write performance of the first performance storage medium is higher than that of the second performance storage medium, and the read / write performance of the second performance storage medium is higher than that of the third performance storage medium.

[0014] A second aspect of this application provides a data temperature assessment device for heavy-haul railway electrical data, comprising: The acquisition module is used to acquire the initial heat assessment values ​​of multiple business data objects in the electrical engineering of heavy-haul railways, as well as the association parameters between any two of the business data objects; The first determining module is used to determine the current popularity value of the target business data object based on the access information, historical popularity value and time decay model of the target business data object in response to the target business data object being accessed. The second determining module is used to determine the heat transmission increment of other business data objects associated with the target business data object based on the association relationship parameters and the current heat value of the target business data object, and to update the current heat value of the other business data objects by combining the historical heat values ​​of the other business data objects and the time decay model; The management module is used to perform hierarchical storage management of the multiple business data objects based on the current popularity value of each of the business data objects.

[0015] A third aspect of this application provides a computer device, including: a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the above methods.

[0016] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any of the preceding claims.

[0017] This application's embodiments obtain initial heat assessment values ​​for several types of data, such as track voltage, switch machines, and track insulation, based on prior experience. Using the Spearman rank correlation coefficient statistical method, correlation analysis is performed on each type of data to construct a data association correlation matrix. The results are then combined with expert scoring to obtain a subjective-objective integrated data correlation matrix. An improved Newton's cooling method is employed to calculate the heat of data objects, while also considering inter-data correlations. Heat calculations are performed on all other data objects to obtain the current state temperature of all data. Combined with data storage space resources, adjustments are made to the data storage medium. This not only achieves accurate perception of the activity level of individual data objects but also constructs a heat-linked ecosystem between data, effectively solving the storage resource mismatch problem caused by isolated assessment perspectives and significantly improving data access efficiency and integrity in complex business scenarios. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram of the data temperature assessment method for heavy-haul railway electrical data provided in the embodiments of this application; Figure 2 This is an overall flowchart of the business data popularity assessment and hierarchical management in the embodiments of this application; Figure 3 This is a flowchart of the initial popularity assessment based on prior knowledge in an embodiment of this application; Figure 4 This is a flowchart illustrating the calculation of the subjective and objective fusion correlation parameters in the embodiments of this application. Figure 5 This is a flowchart of the dynamic update process for heat based on time interval constraints in an embodiment of this application; Figure 6 A schematic diagram of the data temperature assessment device for heavy-haul railway electrical data provided in one embodiment of this application; Figure 7 This is a schematic diagram of a computer device structure provided in one embodiment of this application. Detailed Implementation

[0019] In the process of developing this application, the inventors discovered that the temperature decay of electrical data associated with heavy-haul railways is currently inconsistent. To address this issue, this application provides a method, apparatus, computer equipment, and storage medium for evaluating the temperature of electrical data related to heavy-haul railways.

[0020] The solutions in this application embodiment can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0021] To make the technical solutions and advantages of the embodiments of this application clearer, the exemplary embodiments of this application will be described in further detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not an exhaustive list of all embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0022] Example 1 like Figure 1 and Figure 2 As shown, this embodiment provides a method for evaluating the data temperature of heavy-haul railway electrical data. This method can be executed by a server, processor, or electronic device with data processing capabilities. The method mainly includes the following steps S100-S400: Step S100: Obtain the initial heat assessment values ​​of multiple business data objects in the heavy-haul railway electrical system, as well as the correlation parameters between any two business data objects.

[0023] The initial heat assessment value is a static benchmark value determined based on prior knowledge. It is used to characterize the inherent business value of business data objects when no dynamic access occurs. For example, it can be quantitatively assigned based on dimensions such as data security level, business criticality, or compliance requirements. Business-driven temperature initialization can, for example, rely on the Delphi method to build a full-chain expert team for heavy-load power services. Combined with core business needs, it can standardize the initial temperature scoring of various power service data, accurately anchor the data business value weight, avoid the defects of general methods that are divorced from power service business, and ensure that the assessment conforms to the business rules, security requirements, and operational characteristics of heavy-load power services. The correlation parameter is used to characterize the degree of coupling between any two business data objects, such as the temporal correlation shown in the monitoring data or historical access behavior of the main equipment and auxiliary equipment. In this embodiment, the correlation parameter can be stored in matrix form. The larger the element value in the matrix, the stronger the correlation between the two data objects, and the greater the weight in subsequent heat transmission. This correlation parameter can be pre-calculated and stored in the database, or it can be periodically updated offline and then loaded into memory. This embodiment of the application does not limit this and can be flexibly selected or set according to the actual situation.

[0024] Step S200: In response to the target business data object being accessed, determine the current popularity value of the target business data object based on the access information, historical popularity value and time decay model of the target business data object.

[0025] For example, when the processor detects a read or write request for a specific business data object (i.e., the target business data object), it triggers a popularity update process. Access information includes, but is not limited to, access timestamps, access frequency, and access type (e.g., sequential read, random read); historical popularity values ​​refer to the popularity status of the object at the last evaluation time; a time decay model is used to characterize the natural decrease in data value over time, for example, using an exponential decay function or a linear decay function, so that the popularity of data that has not been accessed for a long time gradually returns to a baseline level. This application embodiment uses the access information of the target business data object, historical popularity values, and the time decay model to reflect the changes in the activity of the target business data object due to its access in real time, ensuring that hot data can be identified in a timely manner.

[0026] Step S300: Based on the association parameters and the current popularity value of the target business data object, determine the popularity transmission increment of other business data objects associated with the target business data object, and update the current popularity value of other business data objects by combining the historical popularity values ​​of other business data objects and the time decay model.

[0027] For example, when a target business data object is accessed and its popularity increases, this popularity is not limited to the target business data object itself, but is "transmitted" to other business data objects associated with it through preset association parameters. The magnitude of the popularity transmission increment is proportional to the association parameters and the access weight of the target business data object; the stronger the association and the more popular the source object, the more popularity is transmitted. Subsequently, this transmission increment is superimposed on the historical popularity decay results of other business data objects to obtain their updated current popularity value. The fundamental reason for introducing this linkage update mechanism is that in complex business scenarios, data is often used in combination in the form of "clusters" or "chains." If only the access indicators of a single piece of data are used for evaluation, it is easy to find that core data has been frequently accessed and promoted to the hot storage layer, while its strongly associated auxiliary data remains in the cold storage layer. This will cause the business to need to cross storage media with different performance when retrieving data, resulting in serious I / O latency and link breakage. Through the association transmission in this step, the synchronous rise and fall of the popularity of associated data is achieved, ensuring the aggregation of strongly associated data at the storage level from a mechanism perspective. The dynamic temperature iteration capability can dynamically calculate data temperature periodically and triggered by an improved Newtonian cooling method, based on factors such as special operating periods, business scenarios, access frequency, and access time of heavy-haul railways. This allows for real-time monitoring of data value benchmarks and updating of data relationships, breaking through the limitations of static layering and adapting to diversified operational needs throughout the entire lifecycle.

[0028] Step S400: Based on the current popularity value of each business data object, perform hierarchical storage management for multiple business data objects.

[0029] After completing the self-updating and related data transmission, all business data objects in the system receive a popularity score representing their current comprehensive value. The processor, based on a pre-defined tiered strategy, migrates or retains data objects with popularity values ​​in different ranges within their corresponding storage tiers. For example, high-popularity data is placed in high-performance storage media (such as SSDs or in-memory databases) to support low-latency access, medium-popularity data is placed in conventional storage media, and low-popularity data is archived to low-cost storage media. This process can be performed periodically in batches or as a real-time migration triggered by a threshold. This refined tiered management, configuring differentiated storage resources for cold, warm, and hot data based on evaluation results—matching high-performance storage to hot data and low-cost archiving to cold data—not only solves the resource waste problem of traditional full-scale homogeneous storage and reduces the cost of massive data storage, but also improves the efficiency of accessing high-value data, supporting the implementation of low-latency, high-reliability scenarios such as real-time fault warnings and intelligent operation and maintenance.

[0030] In the embodiments of this application, steps S100-S400 above obtain initial heat assessment values ​​for several types of data, such as track voltage, switch machine, and track insulation, based on prior experience. Using the Spearman rank correlation coefficient statistical method, correlation analysis is performed on each type of data to construct a correlation matrix of data associations. The results obtained from expert scoring are then fused to obtain a subjective-objective integrated data correlation matrix. An improved Newton cooling method is used to calculate the heat of data objects, while also considering the correlation between data. Heat calculations are performed on all other data objects to obtain the current data temperature of all data. Combined with data storage space resources, adjustments are made to the data storage medium. This not only achieves accurate perception of the activity level of individual data objects but also constructs a heat linkage ecosystem between data, effectively solving the storage resource mismatch problem caused by isolated evaluation perspectives and significantly improving data access efficiency and integrity in complex business scenarios.

[0031] Example 2 Please see Figure 3 and Figure 4 Based on Embodiment 1, this embodiment further refines the specific implementation mechanism for obtaining the association relationship parameters between any two business data objects. In this embodiment, obtaining the association relationship parameters between any two business data objects in step S100 includes the following steps: Step S110: Obtain the subjective association matrix based on prior knowledge. The subjective association matrix contains the subjective association degree between any two business data objects. Step S120: Based on the historical access statistics of multiple business data objects, obtain an objective correlation matrix, which contains the objective correlation degree between any two business data objects. Step S130: Normalize the subjective correlation matrix and the objective correlation matrix, and then perform linear fusion according to preset weights to obtain the fused correlation matrix; Step S140: Determine the association parameters based on the element values ​​in the fusion association matrix.

[0032] This embodiment integrates the subjective experience of expert scoring with the objective laws of Spearman's rank correlation analysis. It uses expert-initialized temperature as a benchmark to control the priority of business value, while also using statistical analysis to uncover implicit correlations between data, calibrating the strength of correlations, correcting subjective biases, and compensating for the shortcomings of purely objective algorithms, thereby improving the comprehensiveness and accuracy of the evaluation. This embodiment constructs a dual verification system of "business knowledge anchoring + mathematical algorithm calibration," avoiding the subjective bias caused by relying solely on expert experience and overcoming the shortcomings of pure data statistics divorced from business semantics. This provides a precise and robust weighting basis for subsequent heat dissipation.

[0033] Please see Figure 3 In an optional embodiment of this application, obtaining the subjective association matrix based on prior knowledge is an iterative process involving the convergence of multi-node scoring and coefficient of variation. Step S110, obtaining the subjective association matrix based on prior knowledge, includes steps S111-S112: Step S111: Obtain the correlation scores between any two business data objects from multiple evaluation nodes, and construct an initial scoring matrix; In this embodiment, the evaluation node can be a domain expert, a senior operations and maintenance personnel, or a business rule engine, and the scoring dimensions cover business logic dependency, frequency of joint fault diagnosis, etc.

[0034] Step S112: Calculate the coefficient of variation of each element in the initial rating matrix. If the coefficient of variation is greater than the preset convergence threshold, trigger re-rating until the coefficient of variation is less than or equal to the preset convergence threshold to obtain the subjective association matrix. For example, regarding the data objects "track voltage" and "switch machine operating current", if some experts believe that the two are strongly correlated (9 points) while others believe that they are weakly correlated (3 points) in the first round of scoring, causing the coefficient of variation to exceed the convergence threshold of 0.2, the system will report the score distribution and trigger the next round of anonymous re-evaluation, forcing extreme opinions to converge to the group consensus, thereby ensuring that the subjective correlation is a stable consensus within the industry, rather than the instantaneous judgment of individual experts, and thus improving the reliability of prior knowledge.

[0035] In an optional embodiment of this application, step S120, obtaining an objective correlation matrix based on the historical access statistics of the multiple business data objects to capture the implicit nonlinear monotonic relationship between the data, specifically includes the following steps: Sort the historical access time series data of any two business data objects, calculate the rank correlation coefficient, and construct an objective correlation matrix based on the absolute value of the rank correlation coefficient.

[0036] The rank correlation coefficient (such as the Spearman coefficient) in this embodiment is not the traditional Pearson correlation coefficient. Access patterns for business data generally exhibit a non-linear, monotonic trend. For example, the access volume of device status data increases in a stepwise manner with the severity of the fault, rather than non-linearly. This embodiment uses the rank correlation coefficient to characterize this hierarchical correlation. Furthermore, this embodiment uses the absolute value of the rank correlation coefficient to eliminate the influence of positive and negative correlation directions. Because in a heat transmission scenario, whether positive or negative, the existence of a strong monotonic relationship implies a close business coupling between the data, and should be considered highly correlated to trigger linked updates.

[0037] This embodiment identifies strongly correlated datasets through correlation analysis such as Spearman, enabling same-layer binding and integrated management. This solves the pain points of scattered and fragmented data retrieval in traditional hierarchical systems, providing a complete and highly correlated dataset for intelligent applications such as power line fault diagnosis and full lifecycle management, thereby improving model accuracy and stability.

[0038] After obtaining the subjective and objective correlation matrices, due to the fundamental differences in their numerical meanings and ranges—the subjective correlation matrix typically has a scale of 1-9, while the objective correlation matrix has coefficients in the range of [-1,1]—they cannot be directly weighted and summed. Therefore, this embodiment maps them to a unified interval of [0,1] to eliminate the dimensional differences. Finally, linear fusion is performed according to preset weights, satisfying the following relationship: The linear fusion according to preset weights yields the fusion correlation matrix:

[0039] in, For the fusion correlation matrix, The normalized subjective association matrix, The normalized objective correlation matrix, The preset subjective weight allocation coefficient is used to adjust the balance between business confidence and data confidence. When the business logic is extremely clear and stable (such as security interlock data), λ can be set to a higher value (such as 0.7-0.8) to make the fusion result more focused on expert experience. When the business logic is ambiguous or in the exploratory stage, but historical access data is abundant, the value of λ can be reduced (such as 0.3-0.4) to facilitate the determination of correlation.

[0040] For example, in a heavy-haul railway electrical system scenario, the data objects "track insulation resistance" and "signal display status" are considered by domain experts to have a strong causal relationship based on circuit principles, with a normalized subjective score of 0.9. However, in recent historical access statistics, due to maintenance habits, these two types of data are rarely accessed simultaneously by the same work order, resulting in a normalized absolute value of only 0.2 for the calculated rank correlation coefficient. If only objective statistics are used, the two will be judged as weakly correlated. Once the track insulation data becomes active, the signal data remains in the cold zone, which may lead to the omission of key information during fault diagnosis. In the fusion model of this embodiment, if λ=0.6, the final fusion correlation degree is 0.6*0.9 + 0.4*0.2 = 0.62. This result retains the potential strong correlation pointed out by experts to avoid missed judgments, while also considering the weakening trend of current actual access patterns to avoid excessive ineffective transmission, achieving a dynamic balance between business theory and actual behavior. It should be understood that the values ​​of the subjective weight allocation coefficient λ and the scoring scale mentioned above are only illustrative examples. In actual applications, they can be adaptively adjusted or configured according to the data characteristics and operation and maintenance requirements of specific business scenarios. This application embodiment does not limit this.

[0041] Example 3 like Figure 5 As shown, this embodiment, based on embodiment 1, further concretizes the functional description of updating the current popularity value of other business data objects into an executable mathematical calculation model. Step S300 above, the updating of the current popularity value of the other business data objects, includes the following steps S310-S320: Step S310: For any of the other business data objects, obtain its historical heat value at the previous evaluation time, its basic observation increment at the current evaluation time, and the heat transmission increment from the target business data object; Step S320: Based on the historical heat value, the basic observation increment, and the heat transmission increment, determine the current heat value of the other business data objects at the current evaluation time; Among them, the historical popularity value decays exponentially with time interval, representing the natural forgetting process of data value over time; the basic observation increment represents the immediate increase in popularity caused by the direct access of the object itself; and the popularity transmission increment is positively correlated with the association parameter and the access weight of the target business data object, representing the popularity spillover effect caused by business coupling between data. Compared with the traditional evaluation method that only focuses on the attributes of a single object, this embodiment introduces an external incentive term, so that data that is not directly accessed but is strongly associated with the accessed object can also receive popularity compensation, thereby mathematically ensuring the synchronization of associated data at the storage level.

[0042] In an optional embodiment of this application, the current popularity value of the other business data objects at the current evaluation time satisfies the following calculation model:

[0043] in, For the first Other business data objects at the current evaluation time The current popularity value, For the previous assessment time Historical popularity value The coefficient of heat decay rate, for and The time interval between This is the basic observation increment. For the target business data object The association parameter between the first other business data object and the first other business data object. This is the access weight function value of the first business data object at the current evaluation time.

[0044] The value ranges and functional forms of the variables in the above calculation model are for illustrative purposes only. In actual deployments, to prevent the heat value from accumulating indefinitely or fluctuating drastically, normalization processing or setting a heat upper limit threshold is usually introduced. This embodiment solidifies the abstract linkage update logic into a three-dimensional mathematical model that includes historical decay, direct increment, and propagation increment. This not only provides a precise quantitative basis for heat assessment but also reserves sufficient room for subsequent parameter optimization for different business characteristics.

[0045] Example 4 like Figure 5 As shown, this embodiment further optimizes the execution efficiency of heat calculation based on embodiment 3, and proposes a branch calculation strategy based on a time interval threshold, specifically including at least the following three cases: The first scenario: When If the heat value is less than or equal to the first preset threshold, skip the calculation of the heat value at the current evaluation time; This design addresses jitter reduction in high-frequency or concurrent access scenarios. In real-world business scenarios, certain hot data may be requested multiple times consecutively within a very short period. If each request triggers a complete hot data update process, it would not only result in a large amount of redundant computation but could also cause I / O bottlenecks due to frequent storage read / write operations. By setting a minimum time threshold 'a', the system treats multiple accesses within a short period as a valid stimulus, triggering an update only when the interval exceeds 'a', thus effectively filtering out invalid high-frequency disturbances. For example, in a heavy-haul railway electrical monitoring scenario, if sensor data is reported at a millisecond frequency, 'a' can be set to 5 to 10 seconds, ensuring timely hot data response while avoiding wasted processor resources.

[0046] The second scenario: when When the current popularity value is greater than the first preset threshold and less than the second preset threshold, the current popularity value of the target business data object is calculated first, and then the current popularity value of all other business data objects is calculated based on the calculation model. when When the value is greater than the first preset threshold 'a' and less than the second preset threshold 'b', the system enters the normal linkage calculation mode. This branch corresponds to the complete three-dimensional calculation model described in Embodiment 3, which simultaneously considers historical decay, its own observed increment, and the heat propagation increment from the target business data object. This is the most accurate state for heat assessment and is suitable for normal business access rhythms. In this mode, the processor first updates the heat of the accessed target business data object, and then traverses the association matrix to propagate the heat increment to all other related business data objects. This complete linkage mechanism ensures that the business coupling relationship between data can be fully expressed under normal access frequency, maintaining the synchronization of related data at the storage level.

[0047] The third scenario: when When the value is greater than or equal to the second preset threshold b, only the time decay term of the plurality of business data objects is calculated; according to Update the current popularity value.

[0048] when When the temperature is greater than or equal to the second preset threshold b, the system enters a hibernation or rapid cooling mode, and only performs the calculation of the time decay term. The generation of the heat transfer increment is predicated on the existence of a triggering event: "the target business data object is accessed." When the time interval... When the second preset threshold b is exceeded, it means that no data objects have been accessed within the current evaluation period, therefore the heat transfer increment of all data is zero. In this case, continuing to traverse the massive relational matrix for multiplication is meaningless, so the system only needs to apply an exponential decay function to the historical heat values ​​cached in memory to complete the update. For example, the second preset threshold b can be set to 12 hours or 24 hours. During the nighttime period when there are no maintenance operations, the system automatically switches to this lightweight mode, reducing the original O(N) time complexity to zero. 2 The complexity of matrix operations is reduced to O(N) vector operations, greatly freeing up computing resources for other background tasks or energy-saving hibernation.

[0049] It should be explained that the specific values ​​of the first preset threshold 'a' and the second preset threshold 'b' are not fixed, but can be adaptively configured or dynamically adjusted according to the load characteristics of the business system, the total amount of data, and the performance of the storage medium. For example, when the system load is high, the value of 'a' can be appropriately increased to reduce the calculation frequency; when the data scale is small, the value of 'b' can be decreased to maintain higher evaluation sensitivity. Furthermore, although this embodiment describes three discrete branch intervals, in other embodiments, continuous functions can be used to smoothly transition the calculation weights of different intervals, or more threshold intervals can be added to adapt to more complex business rhythms. This embodiment, through this time-sharing branching strategy, dynamically adjusts the calculation granularity according to the temporal sparsity of data access, thereby significantly reducing the system's computational overhead while ensuring the real-time performance of the heat assessment. This solves the problem of computational overload or response lag that traditional heat assessment methods easily encounter when dealing with extreme access patterns, achieving an optimal balance between evaluation accuracy and operational efficiency.

[0050] Example 5 like Figure 3 As shown, this embodiment, based on embodiment 1, further refines the specific implementation mechanism for obtaining the initial popularity assessment values ​​of multiple business data objects. In this embodiment, step S100, obtaining the initial popularity assessment values ​​of multiple business data objects in heavy-haul railway electrical engineering, includes the following steps S101-S103: Step S101: Based on the preset multi-dimensional evaluation rules, obtain the anonymous scoring results of multiple evaluation nodes for each of the business data objects; In this embodiment, the multi-dimensional evaluation rules may include, for example, multiple dimensions such as driving safety relevance, fault handling priority, operational decision value, business access frequency, and real-time requirements. Each dimension has a clear scoring scale and weight guide to ensure that different evaluation nodes have a unified reference system when scoring. Evaluation nodes can be domain experts, senior operations and maintenance personnel, or business management personnel. Their scoring process is conducted anonymously to avoid the interference of authority effect or herd mentality on independent judgment. After obtaining the first round of anonymous scoring results, the system does not directly adopt its average value, but instead calculates the coefficient of variation (COP). The COP is the ratio of the standard deviation to the mean. It can eliminate the influence of dimensions and objectively reflect the degree of fluctuation or divergence of a set of data relative to its mean.

[0051] Step S102: Calculate the coefficient of variation of the anonymous scoring result. If the coefficient of variation is greater than the preset convergence threshold, the scoring distribution is fed back and re-scoring is triggered until the coefficient of variation is less than or equal to the preset convergence threshold. Step S103: The average value of the converged anonymous scoring results is determined as the initial popularity assessment value corresponding to the business data object.

[0052] Compared to traditional one-time static assignment or simple arithmetic average, this embodiment eliminates the interference of individual subjective bias and extreme opinions, ensuring that the initial popularity assessment value can truly reflect the stable consensus on data value within the industry, thereby providing a reliable and robust benchmark starting point for subsequent popularity linkage calculations.

[0053] The coefficient of variation (COP) is more sensitive than non-range or variance in capturing the relative dispersion of the scoring distribution, making it particularly suitable for comparing the consensus of data across different score ranges. When the calculated COP exceeds a preset convergence threshold, it indicates significant disagreement among evaluation nodes regarding the value judgment of the business data object, and no effective consensus has yet been reached. At this point, the system triggers a feedback re-scoring mechanism, anonymously providing all evaluation nodes with the statistical distribution of the previous round of scoring (such as median, interquartile range, COP values, etc.), without revealing the specific scoring details of individual nodes. After understanding the overall trend of group opinions, evaluation nodes can re-examine and adjust their scores based on their own business experience. This iterative process of "scoring-statistics-feedback-re-scoring" will continue until the COP of all business data objects falls below the preset convergence threshold. For example, for data objects with clear business logic and mature application scenarios, the convergence threshold can be set more strictly (e.g., 0.1); while for emerging businesses or data types with significant controversy, the threshold can be appropriately relaxed (e.g., 0.2). Only when the discrete coefficients meet the convergence condition will the average score of the final round be solidified as the initial heat assessment value of the object.

[0054] It should be explained that although the initial popularity assessment value acquisition mechanism described in this embodiment and the process of acquiring the subjective correlation matrix in Embodiment 2 both involve expert scoring and convergence testing, they differ fundamentally in their assessment objects and technical objectives. Embodiment 2 focuses on the relative correlation strength between any two business data objects, addressing the question of "how relevant are data A and data B," and its output is an N×N relation matrix; while this embodiment focuses on the absolute value benchmark of a single business data object, addressing the question of "how important is data A itself," and its output is an N-dimensional vector. Embodiment 2 is used to guide the weight allocation during popularity transmission, while this embodiment is used to establish the initial level for popularity decay and accumulation. The two complement each other and together constitute a complete popularity assessment system.

[0055] Through the aforementioned iterative convergence mechanism, this embodiment effectively avoids the problem of initial value distortion caused by individual expert cognitive biases or information asymmetry. For example, if a certain type of monitoring data is misjudged as extremely valuable (10 points) by a few experts but considered average (4 points) by most experts, a simple average might yield a slightly above-average result of 7 points. This would cause the data to be incorrectly placed in the hot storage layer during system startup, consuming valuable high-performance resources. However, under the mechanism of this embodiment, because the first-round dispersion coefficient significantly exceeds the standard, the system will forcibly trigger multiple rounds of feedback and correction, prompting extreme opinions to revert to group rationality. The final initial heat assessment value will be closer to the actual business needs. This initialization method based on statistical consensus not only improves the cold-start accuracy of the heat assessment model but also provides a stable anchor point for subsequent dynamic heat calculations, preventing frequent oscillations in the storage layer caused by benchmark drift, and ensuring the effectiveness and economy of the tiered storage management strategy from the source. It should be understood that the above-mentioned formulas for calculating the coefficient of variation, the specific values ​​of the convergence threshold, and the presentation format of the feedback information are all exemplary illustrations. In practical applications, they can be adaptively adjusted according to the scale of the evaluation node, the maturity of the business domain, and the sensitivity of the system to the initial accuracy. This application embodiment does not limit these aspects.

[0056] Example 6 In this embodiment, the aforementioned general business data popularity assessment method is specifically applied to the storage management scenario of heavy-haul railway electrical monitoring data. Specifically, the business data objects include heavy-haul railway electrical monitoring data, which includes at least one of track voltage data, switch machine status data, track insulation data, signal operating current data, and environmental status data.

[0057] Heavy-haul railway electrical systems are characterized by diverse equipment types, high data coupling, and strong fault correlation. Different monitoring data have vastly different real-time requirements and access patterns for ensuring train operation safety. For example, track voltage data directly reflects the working status of track circuits and is the core basis for judging faults endangering train operation safety, such as red light bands and faulty circuits. It is accessed very frequently and is extremely sensitive to delays. Switch machine status data records the operating current and indication gap information during turnout switching, and is key evidence for analyzing mechanical faults such as turnout jamming and improper contact adjustment. It is usually accessed centrally during fault investigation or periodic maintenance. Track insulation data is used to monitor the electrical performance of rail insulation joints and is fundamental data for preventative maintenance. Signal operating current data reflects the load status of the signal display mechanism. Environmental status data (such as temperature, humidity, wind speed, and rainfall) mainly serve as auxiliary background information for fault analysis, with relatively low access frequency but requiring long-term storage to meet traceability compliance requirements. By incorporating these heterogeneous data into a unified heat assessment system, the rigid management model of fixed partitions based on data type can be broken, enabling dynamic resource allocation based on actual business value.

[0058] For the aforementioned business data objects, a hierarchical storage management system is implemented for multiple business data objects, including: dividing multiple business data objects into a hot data layer, a warm data layer, and a cold data layer based on a preset temperature scoring range; storing the hot data layer data in a first performance storage medium, storing the warm data layer data in a second performance storage medium, and storing the cold data layer data in a third performance storage medium, wherein the read / write performance of the first performance storage medium is higher than that of the second performance storage medium, and the read / write performance of the second performance storage medium is higher than that of the third performance storage medium.

[0059] The preset temperature rating range can be configured according to actual operation and maintenance needs. As a preferred implementation, data with a heat score between 80 and 100 can be designated as the hot data layer, data between 40 and 79 as the warm data layer, and data between 0 and 39 as the cold data layer. In terms of storage media selection, the first-performance storage medium can be NVMe SSDs or high-performance DRAM caches to meet the requirements of millisecond-level random reads and high-concurrency writes for hot data; the second-performance storage medium can be enterprise-grade SAS or SATA hard disk arrays, providing large storage capacity and low unit cost while ensuring a certain throughput; the third-performance storage medium can be tape libraries, Blu-ray storage, or low-cost object storage, dedicated to the long-term archiving and compliant retention of cold data. This three-tier storage architecture not only achieves gradient optimization of storage costs, but more importantly, provides a physical carrier for the heat-linked mechanism, enabling data to migrate smoothly between media of different performance levels according to changes in business value.

[0060] Example 7 To more intuitively demonstrate the technical advantages of the embodiments of this application in the heavy-haul railway electrical system scenario, this embodiment takes a typical joint fault diagnosis scenario as an example to specifically explain the data temperature assessment method for heavy-haul railway electrical system data provided in the embodiments of this application.

[0061] I. Obtaining the initial heat assessment values ​​of multiple business data objects in heavy-haul railway electrical systems This embodiment mainly divides the Delphi method heat initialization assessment process for heavy-haul railway electrical engineering into three stages: assessment preparation stage, assessment execution stage, and heat consolidation stage. Based on the core principles of the Delphi method—anonymity, multi-round iteration, and consensus convergence—multiple rounds of standardized anonymous scoring and opinion iteration by experts in the heavy-haul railway electrical engineering field are used to quantitatively assign initial heat values ​​to all heavy-haul railway electrical engineering data. This provides a benchmark basis that aligns with the core business for subsequent data stratification calibration. The specific work content of each stage is as follows: 1) Assessment Preparation Phase: Establish assessment rules for business adaptation: Develop quantitative scoring dimensions and scoring scales that fit heavy-load power services. The core assessment dimensions are set as follows: correlation with driving safety, priority of fault handling, value of operation and maintenance decisions, frequency of business access, and real-time requirements. At the same time, the relationship between the heat score range and the level is clarified. For example: high temperature heat data (80-100 points), medium temperature heat data (40-79 points), and low temperature cold data (0-39 points).

[0062] Establish an expert database for the field of power supply business: The experts cover the entire chain of front-line operation and maintenance, system design, fault diagnosis and intelligent application of heavy-haul power supply, including core technical backbones, railway bureau power supply management experts, system R&D experts and other multi-dimensional subjects, to ensure that the experts' opinions cover all business scenarios of heavy-haul power supply.

[0063] Identify and compile a list of data to be evaluated: Complete the comprehensive review of the heavy-haul railway electrical data system, standardize the classification according to business attributes, clarify the business boundaries and application scenarios of each type of data, and compile a list of data to be evaluated.

[0064] 2) Evaluation and Implementation Phase The first round of evaluation was conducted by anonymously distributing evaluation questionnaires and basic data to all experts. Each expert then independently scored each type of data from multiple dimensions based on their own business experience, resulting in the first round of expert scoring results.

[0065] Scoring results statistics: The scoring results of the first round are statistically analyzed, and the median, mean, interquartile range and coefficient of variation of the scores for each data category are calculated to form an anonymous feedback report, which is then distributed to all experts.

[0066] Multiple rounds of iterative feedback: Based on the overall feedback from the previous round of scoring, experts, combined with their own business experience and referring to the coefficient of variation of each data point, adjust and correct the scoring results. The process of "result statistics - anonymous feedback - iterative scoring" is repeated until the coefficient of variation of all data items is lower than the preset convergence threshold (usually ≤0.1), ensuring that the experts' opinions reach an industry consensus.

[0067] 3) Heat curing stage Determine the initial popularity assessment value: Calculate the average score of each type of electrical data based on the final expert scoring results after convergence. This score is the initial popularity assessment value of the corresponding data.

[0068] Construct a heat index benchmark library: Dynamically adjust the initial heat index benchmark library for heavy-haul railway electrical data, record the initial heat index assessment value for each type of data, and provide a business benchmark for subsequent heat index calibration, dynamic adjustment and hierarchical storage based on Spearman correlation analysis.

[0069] II. Calculation of Association Parameters (1) Obtaining the subjective association matrix based on prior knowledge: 1) Select A panel of experts was formed, and each expert scored the data based on the interrelationships between the N sets of data in the field of railway electrical engineering.

[0070] No. The expert rating matrix provided by the experts is as follows:

[0071] in Indicates the first Experts on the data With data The score of correlation between them This rating is used to measure data. With data The data relevance is assessed using a 1-9 scale, where 1 indicates a weak relationship or minor influence between the two sets of data, and 9 indicates a very strong correlation or influence between one set of data and the other set.

[0072] 2) To The initial rating matrix is ​​obtained by aggregating the rating matrices provided by renowned experts:

[0073] in:

[0074] 3) To ensure the reliability of expert opinions, a consistency check is performed on each matrix element. (Regarding the data...) With data Calculate the standard deviation of expert ratings :

[0075] (4) Calculate the coefficient of variation of the expert rating matrix for each round:

[0076] in The coefficient of variation represents the opinions of various experts on the data. With data Differences in correlation scores between them.

[0077] (5) If If the value exceeds the preset threshold, it indicates a significant disagreement among experts regarding the relevance of the data, requiring iterative iterations of steps 1) to 4) until... If the value is below the preset threshold, it indicates that expert opinions are stabilizing. This is the subjective association matrix of the final subjective data.

[0078] (2) Obtain the objective correlation matrix 1) Based on N sets of original data, calculate the Spearman rank correlation coefficient between any two sets of data. and The number of data points is 1. And it corresponds to the timestamp in the data. The first data point taken from the two sets of random variables... Each value is used , express, .right , Perform both descending and ascending sorting simultaneously to obtain two data ranking sets. , ,in , They are respectively exist Sort number and exist The sorting number in the sequence.

[0079] For any two data and Its Spearman rank correlation coefficient is:

[0080] Spearman correlation coefficient This represents the direction and strength of the monotonic correlation between two variables, where and These represent the data ranking sets respectively. , The average value.

[0081]

[0082] in, Indicates a perfect positive correlation. Indicates a completely negative correlation. This indicates that there is no monotonic correlation.

[0083] 2) For N sets of data, the following objective correlation matrix of Spearman objective data can be obtained:

[0084] (3) Determine the fusion correlation matrix The subjective correlation matrix of expert subjective data obtained by the Delphi method The objective correlation matrix of objective data obtained by the Spearman method Because of differences in their numerical meanings and ranges, the two cannot be directly weighted and summed. The elements in the subjective association matrix represent experts' subjective ratings of the correlation or influence between the data, while the elements in the Spearman objective association matrix represent the monotonic correlation direction and strength between the two sets of data. To fuse the two types of matrices at the same scale, both the subjective and objective association matrices need to be normalized.

[0085] 1) First, regarding After normalization, the subjective correlation matrix is ​​obtained. :

[0086] in:

[0087]

[0088] 2) Similarly, for the Spearman objective correlation matrix... Normalization is performed. Since the Spearman correlation coefficient ranges from -1 to 1, its sign indicates the direction of the correlation, and the absolute value indicates the strength of the correlation. If we only focus on the strength of the correlation between the data, then after taking the absolute value of the Spearman correlation coefficient... After normalization, the Spearman objective correlation matrix is ​​obtained. :

[0089] in:

[0090]

[0091]

[0092] 3) After obtaining the subjective association matrix And Spearman's objective correlation matrix Then, a linear weighted method was used to fuse subjective and objective data to obtain the fusion correlation matrix. :

[0093] in, This is the subjective fusion coefficient, and its value range is:

[0094] when A larger value indicates that the fusion result places more emphasis on subjective expert opinions; when... A smaller value indicates that the fusion result focuses more on the objective correlation between data.

[0095] final or For data With data The correlation.

[0096] III. Calculation of Current Popularity Value By combining access time, access frequency, and the correlation between data, this embodiment constructs a railway data heat calculation method that combines "triggering and periodicity".

[0097] Assuming the electrical engineering data includes data , , ... , The overall principle of this embodiment is "real-time calculation and synchronous update," meaning that when data is accessed, the popularity of all data at that moment is calculated synchronously. To ensure the efficiency and availability of popularity calculation, this embodiment is based on the time interval for calculating popularity. , , The time for the m-th heat calculation is defined as follows: Minimum value of the first preset threshold Maximum value second preset threshold Therefore, popularity calculation is divided into the following three cases: (1) When That is, the moment when the data is accessed. The time since the last heat calculation If the value is less than the minimum, then this moment is not calculated. The popularity.

[0098] (2) When That is, the moment when the data is accessed. time interval Greater than the minimum value And less than the maximum value Then, it is necessary to calculate all data at time [time]. The popularity of the accessed data object needs to be calculated first to determine the relationships between the data objects. Then, the popularity of all other data objects needs to be calculated.

[0099] 1) Assuming data This refers to the data object being accessed. First, based on Newton's law of cooling and combined with access frequency weighting, the data is calculated. At Real-time popularity based on the number of visits:

[0100] in The coefficient of heat decay rate, For data-based Weight function value of historical visits:

[0101] in, In order to be in Time data Total visits The moment of initial heat.

[0102] 2) Combining data The visitor weighting function and the correlation coefficient between data are used to calculate the remaining data. , , ... Real-time popularity value at the same time. Based on data. For example, data In this scenario The heat value at that time was:

[0103] in For data With data Data relevance parameters.

[0104] 3) And so on, data In this scenario The temperature at that time was:

[0105] (3) When That is, if the time interval between the current time and the last time the heat was calculated is greater than the maximum value. Then, it is necessary to do so at the current moment. Calculate all data , , ... The popularity. At this time, the data... The popularity is:

[0106] IV. Data Cold, Warm, and Hot Encapsulation Storage Management Hot / Cold Temperature Classification: Based on the initially preset scoring range, hot / cold temperature classification is carried out to clearly define the popularity level of each data type. The preset scoring range can be adjusted later according to business characteristics and storage media requirements.

[0107] It should be understood that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order constraint on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the diagram may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0108] Please see Figure 6 One embodiment of this application provides a data temperature assessment device 600 for heavy-haul railway electrical data, comprising: The acquisition module 610 is used to acquire the initial heat assessment value of multiple business data objects in the electrical system of heavy-haul railway, as well as the association parameters between any two of the business data objects; The first determining module 620 is used to determine the current popularity value of the target business data object based on the access information, historical popularity value and time decay model of the target business data object in response to the target business data object being accessed. The second determining module 630 is used to determine the heat transmission increment of other business data objects associated with the target business data object based on the association relationship parameters and the current heat value of the target business data object, and to update the current heat value of the other business data objects by combining the historical heat values ​​of the other business data objects and the time decay model; The management module 640 is used to perform hierarchical storage management of the multiple business data objects based on the current popularity value of each of the business data objects.

[0109] In an optional embodiment of this application, the acquisition module 610 is specifically used to: acquire a subjective association matrix based on prior knowledge, the subjective association matrix containing the subjective association degree between any two of the business data objects; acquire an objective association matrix based on historical access statistics of the plurality of business data objects, the objective association matrix containing the objective association degree between any two of the business data objects; normalize the subjective association matrix and the objective association matrix, and linearly fuse them according to preset weights to obtain a fused association matrix; and determine the association relationship parameters based on the element values ​​in the fused association matrix.

[0110] In an optional embodiment of this application, the acquisition module 610 is specifically used to: acquire correlation scores between any two business data objects from multiple evaluation nodes, and construct an initial scoring matrix; calculate the coefficient of variation of each element in the initial scoring matrix; if the coefficient of variation is greater than a preset convergence threshold, trigger re-scoring until the coefficient of variation is less than or equal to the preset convergence threshold, thereby obtaining the subjective correlation matrix; and / or sort the historical access time-series data of any two business data objects, calculate the rank correlation coefficient, and construct the objective correlation matrix based on the absolute value of the rank correlation coefficient; and perform linear fusion according to preset weights to obtain a fused correlation matrix.

[0111] in, For the fusion correlation matrix, The normalized subjective association matrix, The normalized objective correlation matrix, The preset subjective weight allocation coefficients.

[0112] In an optional embodiment of this application, the first determining module 620 is specifically configured to, for any of the other business data objects, obtain its historical popularity value at the previous evaluation time, its basic observation increment at the current evaluation time, and the popularity transmission increment from the target business data object; and determine the current popularity value of the other business data object at the current evaluation time based on the historical popularity value, the basic observation increment, and the popularity transmission increment; wherein the historical popularity value decays exponentially with time intervals, and the popularity transmission increment is positively correlated with the association parameter and the access weight of the target business data object.

[0113] In an optional embodiment of this application, the current popularity value of the other business data objects at the current evaluation time satisfies the following calculation model:

[0114] in, For the first Other business data objects at the current evaluation time The current popularity value, For the previous assessment time Historical popularity value The coefficient of heat decay rate, for and The time interval between This is the basic observation increment. For the target business data object The association parameter between the first other business data object and the first other business data object. This is the access weight function value of the first business data object at the current evaluation time.

[0115] In an optional embodiment of this application, the first determining module 620 is further configured to, based on the time interval Execute branch calculation strategy: when When the heat value is less than or equal to the first preset threshold, skip the calculation of the heat value at the current evaluation time; when When the value is greater than the first preset threshold and less than the second preset threshold, the current popularity value of the target business data object is first calculated, and then the current popularity value of all other business data objects is calculated based on the calculation model; when When the threshold value is greater than or equal to the second preset threshold, only the time decay term of the plurality of business data objects is calculated; according to Update the current popularity value.

[0116] In an optional embodiment of this application, the acquisition module 610 is specifically used to: acquire anonymous scoring results of multiple evaluation nodes for each of the business data objects based on preset multi-dimensional evaluation rules; calculate the coefficient of variation of the anonymous scoring results; if the coefficient of variation is greater than a preset convergence threshold, feed back the scoring distribution and trigger re-scoring until the coefficient of variation is less than or equal to the preset convergence threshold; and determine the average value of the converged anonymous scoring results as the initial popularity evaluation value of the corresponding business data object.

[0117] In an optional embodiment of this application, the business data objects include heavy-haul railway electrical monitoring data, which includes at least one of track voltage data, switch machine status data, track insulation data, signal operating current data, and environmental status data. Specifically, the management module 640 is used to divide the multiple business data objects into a hot data layer, a warm data layer, and a cold data layer based on a preset temperature scoring range; store the hot data layer data in a first performance storage medium, store the warm data layer data in a second performance storage medium, and store the cold data layer data in a third performance storage medium, wherein the read / write performance of the first performance storage medium is higher than that of the second performance storage medium, and the read / write performance of the second performance storage medium is higher than that of the third performance storage medium.

[0118] Specific limitations regarding the data temperature assessment device 600 for heavy-haul railway electrical data can be found in the above-described limitations on the data temperature assessment method for heavy-haul railway electrical data, and will not be repeated here. Each module in the data temperature assessment device 600 for heavy-haul railway electrical data can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0119] In one embodiment, a computer device is provided, the internal structure of which can be as shown in the figure. Figure 7 As shown. The computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores data. The network interface communicates with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method for assessing the data temperature of heavy-haul railway electrical data. It includes: a memory and a processor; the memory stores a computer program; and the processor executes the computer program to implement any step in the above-described method for assessing the data temperature of heavy-haul railway electrical data.

[0120] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, can perform any step in the data temperature assessment method for heavy-haul railway electrical data as described above.

[0121] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0122] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0123] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0124] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0125] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0126] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for evaluating the data temperature of electrical signaling data in heavy-haul railways, characterized in that, include: Obtain the initial heat assessment values ​​of multiple business data objects in the electrical system of heavy-haul railways, as well as the association parameters between any two of the business data objects; In response to the access of a target business data object, the current popularity value of the target business data object is determined based on the access information, historical popularity value, and time decay model of the target business data object; Based on the association parameters and the current popularity value of the target business data object, determine the popularity transmission increment of other business data objects associated with the target business data object, and update the current popularity value of the other business data objects by combining the historical popularity values ​​of the other business data objects and the time decay model; Based on the current popularity value of each of the business data objects, the multiple business data objects are stored and managed in a hierarchical manner.

2. The method according to claim 1, characterized in that, The step of obtaining the association parameter between any two of the business data objects includes: A subjective association matrix is ​​obtained based on prior knowledge, and the subjective association matrix contains the subjective association degree between any two of the business data objects. Based on the historical access statistics of the multiple business data objects, an objective correlation matrix is ​​obtained, which contains the objective correlation degree between any two of the business data objects. The subjective correlation matrix and the objective correlation matrix are normalized and then linearly fused according to preset weights to obtain a fused correlation matrix. The association parameters are determined based on the element values ​​in the fusion association matrix.

3. The method according to claim 2, characterized in that, The method of obtaining the subjective association matrix based on prior knowledge includes: Obtain the correlation scores between any two business data objects from multiple evaluation nodes, and construct an initial score matrix; Calculate the coefficient of variation for each element in the initial rating matrix. If the coefficient of variation is greater than a preset convergence threshold, trigger a re-rating process until the coefficient of variation is less than or equal to the preset convergence threshold, thus obtaining the subjective association matrix; and / or, The process of obtaining an objective correlation matrix based on historical access statistics of the multiple business data objects includes: Sort the historical access time-series data of any two of the aforementioned business data objects, calculate the rank correlation coefficient, and construct the objective correlation matrix based on the absolute value of the rank correlation coefficient; and / or, The linear fusion according to preset weights yields the fusion correlation matrix: ； in, For the fusion correlation matrix, The normalized subjective association matrix, The normalized objective correlation matrix, The preset subjective weight allocation coefficients.

4. The method according to claim 1, characterized in that, Updating the current popularity value of the other business data objects includes: For any of the other business data objects, obtain its historical popularity value at the previous evaluation time, its basic observation increment at the current evaluation time, and the popularity transmission increment from the target business data object; Based on the historical popularity value, the basic observation increment, and the popularity transmission increment, determine the current popularity value of the other business data objects at the current evaluation time; The historical popularity value decays exponentially with time intervals, and the popularity transmission increment is positively correlated with the correlation parameter and the access weight of the target business data object.

5. The method according to claim 4, characterized in that, The current popularity value of the other business data objects at the current evaluation time satisfies the following calculation model: ； in, For the first Other business data objects at the current evaluation time The current popularity value, For the previous assessment time Historical popularity value The coefficient of heat decay rate, for and The time interval between This is the basic observation increment. For the target business data object The association parameter between the first other business data object and the first other business data object. This is the access weight function value of the first business data object at the current evaluation time.

6. The method according to claim 5, characterized in that, The method also includes based on the time interval. Execute branch calculation strategy: when If the heat value is less than or equal to the first preset threshold, skip the calculation of the heat value at the current evaluation time; when When the current popularity value is greater than the first preset threshold and less than the second preset threshold, the current popularity value of the target business data object is calculated first, and then the current popularity value of all other business data objects is calculated based on the calculation model. when When the threshold is greater than or equal to the second preset threshold, only the time decay term of the plurality of business data objects is calculated; according to Update the current popularity value.

7. The method according to claim 1, characterized in that, The process of obtaining the initial heat assessment values ​​of multiple business data objects in heavy-haul railway electrical systems includes: Based on preset multi-dimensional evaluation rules, anonymous scoring results of multiple evaluation nodes for each of the business data objects are obtained. Calculate the coefficient of variation of the anonymous scoring results. If the coefficient of variation is greater than a preset convergence threshold, then the scoring distribution is fed back and a re-scoring is triggered until the coefficient of variation is less than or equal to the preset convergence threshold. The average value of the converged anonymous scoring results is determined as the initial popularity assessment value for the corresponding business data object.

8. The method according to claim 1, characterized in that, The business data objects include heavy-haul railway electrical monitoring data, which includes at least one of track voltage data, switch machine status data, track insulation data, signal machine operating current data, and environmental status data. The hierarchical storage management of the multiple business data objects includes: Based on the preset temperature scoring range, the multiple business data objects are divided into a hot data layer, a warm data layer, and a cold data layer. The hot data layer data is stored in a first performance storage medium, the warm data layer data is stored in a second performance storage medium, and the cold data layer data is stored in a third performance storage medium, wherein the read / write performance of the first performance storage medium is higher than that of the second performance storage medium, and the read / write performance of the second performance storage medium is higher than that of the third performance storage medium.

9. A data temperature assessment device for heavy-haul railway electrical data, characterized in that, include: The acquisition module is used to acquire the initial heat assessment values ​​of multiple business data objects in the electrical engineering of heavy-haul railways, as well as the association parameters between any two of the business data objects; The first determining module is used to determine the current popularity value of the target business data object based on the access information, historical popularity value and time decay model of the target business data object in response to the target business data object being accessed. The second determining module is used to determine the heat transmission increment of other business data objects associated with the target business data object based on the association relationship parameters and the current heat value of the target business data object, and to update the current heat value of the other business data objects by combining the historical heat values ​​of the other business data objects and the time decay model; The management module is used to perform hierarchical storage management of the multiple business data objects based on the current popularity value of each of the business data objects.

10. A computer device, comprising: A memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method according to any one of claims 1 to 8.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 8.

Images