Oil and gas inventory dynamic optimization method based on big data

Abstract

The present application relates to the technical field of inventory management, in particular to an oil and gas inventory dynamic optimization method based on big data, which collects oil and gas inventory sequences, divides frequency bands to identify abnormalities, extracts mutual information to screen associated category combinations, calculates inventory proportion changes to obtain difference data, analyzes fluctuation characteristics to form characteristic data, and generates optimization suggestions. The present application extracts inventory data in continuous time sequences and divides frequency bands to identify inventory abnormal variation intervals, identifies associated combinations according to the mutual information strength of inventory sequences between categories, and further combines the inventory proportion change amplitude and fluctuation frequency characteristics to construct the coupling analysis between the inventory change trend and the fluctuation behavior of oil and gas categories, realize the quantitative measurement of the continuity of the inventory change direction and the fluctuation duration, form the multi-factor sorting basis for the inventory proportion change and the inventory fluctuation significance, and effectively improve the pertinence of the inventory adjustment suggestions and the rationality of the priority sorting.

Description

Technical Field

[0001] This invention relates to the field of inventory management technology, and in particular to a method for dynamic optimization of oil and gas inventory based on big data. Background Technology

[0002] The field of oil and gas inventory management technology involves the collection, analysis, prediction, and regulation of inventory information generated during the transportation, storage, supply, and consumption of energy products such as oil and natural gas. This includes inventory level monitoring, demand forecasting, replenishment strategy formulation, inventory distribution optimization, and risk assessment.

[0003] Among them, the dynamic optimization method for oil and gas inventory refers to the management and adjustment of oil and gas inventory changes over time in order to balance supply and demand and inventory costs. Under the influence of multiple factors such as fluctuations in oil and gas consumption, limited supply cycles, and storage constraints, a reasonable inventory scheduling plan is dynamically formulated.

[0004] Because routine management and adjustments are based solely on inventory changes over time, without a deep analysis of the fluctuation characteristics at different frequency bands within the inventory sequence, it becomes impossible to identify the specific frequency bands and correlations of abnormal inventory changes in a timely manner when oil and gas consumption fluctuates dramatically or supply changes cyclically. This results in a lack of timeliness and precision in inventory scheduling strategies. When faced with interconnected or synchronous trends between oil and gas categories, conventional methods struggle to identify potential inventory impact mechanisms, leading to problems such as delayed inventory warnings, resource allocation imbalances, and slow supply responses. In particular, in scenarios where inventory changes are frequent and recurring in short cycles, conventional technologies lack multi-dimensional fluctuation identification methods and are unable to provide effective adjustment guidance, ultimately affecting the efficiency of inventory management in responding to sudden fluctuations and the accuracy of adjustment operations. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a dynamic optimization method for oil and gas inventory based on big data.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a dynamic optimization method for oil and gas inventory based on big data, comprising the following steps:

[0007] S1: Collect the inventory sequence of each oil and gas category within the monitoring period under the big data environment and divide it into multiple frequency bands according to the time window. Based on the differences in inventory distribution between each frequency band, filter out frequency bands with abnormal inventory changes.

[0008] S2: Obtain commodity inventory information for all oil and gas categories within the abnormal inventory change frequency band, calculate the mutual information of each pair of oil and gas categories in the current frequency band, and filter out the associated oil and gas category pairs that exceed the mutual information threshold.

[0009] S3: Calculate the current proportion of each of the associated oil and gas product categories in the total inventory, compare it with the proportion of the previous frequency band, mark the inventory change of each oil and gas product category, and obtain inventory proportion difference marker data.

[0010] S4: Analyze the fluctuation pattern of inventory changes in each abnormal frequency band of the inventory change abnormality band, extract the fluctuation features that exceed the fluctuation threshold, and integrate them into inventory fluctuation feature data;

[0011] S5: Obtain the oil and gas categories corresponding to the inventory ratio difference marker data and inventory fluctuation characteristic data, analyze the continuity of the inventory change direction and the duration of fluctuation of the target oil and gas category, establish an adjustment ranking, and generate inventory optimization adjustment suggestions.

[0012] The present invention improves upon the following: the abnormal inventory change frequency band includes a frequency band number, an associated oil and gas category identifier, and a change range; the associated oil and gas category pair combination includes an oil and gas category pair number, a synchronization relationship strength, and a statistical time period; the inventory ratio difference marker data includes the direction of inventory ratio change, the magnitude of ratio change, and a reference value for the total ratio of the combination; the inventory fluctuation characteristic data includes frequency characteristics, duration range, and fluctuation component number; and the inventory optimization and adjustment suggestions include the adjustment direction, sorting priority, and corresponding oil and gas category.

[0013] The present invention is improved in that step S1 is specifically as follows:

[0014] S101: Collect the inventory sequence of each oil and gas category within the monitoring period under the big data environment, obtain the original inventory records of all oil and gas categories in the continuous time dimension, and obtain the original inventory sequence of oil and gas categories.

[0015] S102: Divide the original inventory sequence of each oil and gas category into fixed time windows, extract the inventory value of each oil and gas category in each time window, and generate the time frequency band division result of oil and gas categories;

[0016] S103: Based on the time frequency band division results of the oil and gas categories, calculate the difference in inventory value distribution between adjacent time frequency bands for each oil and gas category, compare it with the set inventory change threshold, identify the time frequency bands where the inventory change exceeds the threshold, and obtain the abnormal inventory change frequency bands.

[0017] The present invention is improved in that step S2 is specifically as follows:

[0018] S201: Obtain each time band marked in the abnormal inventory change frequency band and the corresponding oil and gas category, extract the commodity inventory information of all oil and gas categories in the current frequency band, and construct the oil and gas category inventory sequence in the abnormal inventory frequency band.

[0019] S202: Based on the oil and gas category inventory sequence within the abnormal inventory frequency band, construct oil and gas category pairs according to the pairwise combination method within the current frequency band, extract the inventory sequence of each oil and gas category pair within the corresponding frequency band, and calculate the mutual information value of the oil and gas category pairs.

[0020] S203: For each pair of oil and gas categories, compare the mutual information value of the oil and gas category pair with the mutual information threshold in the current frequency band, filter out oil and gas category pairs with mutual information values ​​greater than the mutual information threshold, record them as related combinations, and obtain related oil and gas category pair combinations.

[0021] The present invention is improved in that step S3 is specifically as follows:

[0022] S301: Obtain the associated oil and gas category pairs, extract the inventory values ​​of all associated oil and gas categories in the current frequency band, and calculate the proportion of the total inventory of each oil and gas category in the total inventory in the current frequency band, generating the inventory proportion data of oil and gas categories in the current frequency band.

[0023] S302: Based on the current frequency band inventory ratio data of the oil and gas category, extract the total inventory of the corresponding oil and gas category in the previous frequency band, calculate the inventory ratio of the previous frequency band, call the ratio data of the current frequency band and the previous frequency band to compare the difference, and obtain the change range data of the inventory ratio of the oil and gas category.

[0024] S303: Based on the data on the change in the inventory ratio of the oil and gas categories, mark the direction and magnitude of inventory change for each oil and gas category, record the positive or negative value of the difference and retain the inventory value, identify the proportion difference of each oil and gas category during the frequency band change process, and obtain inventory ratio difference marking data.

[0025] The present invention is improved in that step S4 is specifically as follows:

[0026] S401: Obtain all time frequency bands and corresponding oil and gas categories marked in the abnormal inventory change frequency bands, extract the inventory time series of each oil and gas category in each abnormal frequency band, and generate an abnormal frequency band inventory sequence for oil and gas categories.

[0027] S402: Based on the abnormal frequency band inventory sequence of the oil and gas category, perform sequence-by-sequence decomposition on each inventory sequence using the empirical mode decomposition algorithm, extract all intrinsic mode function components, and calculate the fluctuation frequency and time distribution of each component to obtain the fluctuation frequency of the oil and gas category inventory components.

[0028] S403: Based on the fluctuation frequency of the oil and gas product inventory components, compare the fluctuation frequency value of each component with the set fluctuation frequency threshold item by item, mark the components whose frequency value exceeds the fluctuation frequency threshold, extract the corresponding continuous segment and change pattern characteristics, and obtain inventory fluctuation characteristic data.

[0029] The present invention is improved in that the intrinsic mode function components are multiple independent fluctuation components decomposed from the inventory time series of oil and gas products in abnormal frequency bands, and each component represents the inventory change process with corresponding frequency and amplitude level.

[0030] The present invention is improved in that step S5 is specifically as follows:

[0031] S501: Obtain the corresponding oil and gas categories in the inventory ratio difference marker data and the inventory fluctuation feature data, extract the inventory change trend segment of each oil and gas category in the current frequency band, identify the continuity status of the inventory change direction based on the continuous rise and continuous fall direction change interval in the inventory sequence, and generate the inventory direction continuity information of oil and gas categories.

[0032] S502: Based on the inventory direction continuity information of the oil and gas categories, extract the continuous segment of the corresponding oil and gas category in the inventory fluctuation characteristics, record the time span of the component corresponding to the frequency threshold, and pair it with the direction continuity interval to determine the degree of overlap between the change trend and the fluctuation behavior, and obtain the inventory trend continuity information of the oil and gas category.

[0033] S503: Based on the information on the continuity of oil and gas product inventory trends, the duration of the direction of inventory change and the duration of the fluctuation frequency component are used as sorting criteria. The sorting priority is set, and the products are sorted according to the strength of continuity and the duration of fluctuation. The adjustment order and direction of each oil and gas product are marked, and inventory optimization adjustment suggestions are obtained.

[0034] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0035] This invention extracts inventory data from continuous time series and divides it into frequency bands to identify abnormal inventory fluctuation intervals. It then identifies associated combinations based on the mutual information strength of inventory sequences between product categories. Furthermore, by combining the magnitude of inventory percentage changes and fluctuation frequency characteristics, it constructs a coupled analysis between the inventory change trend and fluctuation behavior of oil and gas products. This enables a quantitative measurement of the continuity of inventory change direction and the duration of fluctuations, forming a multi-factor ranking basis for inventory percentage changes and the significance of inventory fluctuations. This effectively improves the pertinence of inventory adjustment suggestions and the rationality of priority ranking, avoiding resource waste caused by blind adjustments in inventory scheduling. Simultaneously, it enhances the accuracy of inventory change identification through multi-source data fusion and correlation analysis, ensuring the effectiveness and adaptability of inventory optimization strategies in complex inventory environments, thereby promoting refined and proactive control of oil and gas inventory dynamic management. Attached Figure Description

[0036] Figure 1 This is a flowchart of the method of the present invention;

[0037] Figure 2 This is a detailed flowchart of step S1 of the present invention;

[0038] Figure 3 This is a detailed flowchart of step S2 of the present invention;

[0039] Figure 4 This is a detailed flowchart of step S3 of the present invention;

[0040] Figure 5 This is a detailed flowchart of step S4 of the present invention;

[0041] Figure 6 This is a detailed flowchart of step S5 of the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0043] Please see Figure 1 This invention provides a technical solution: a dynamic optimization method for oil and gas inventory based on big data, comprising the following steps:

[0044] S1: Collect the inventory sequence of each oil and gas category within the monitoring period under the big data environment and divide it into multiple frequency bands according to the time window. Based on the differences in inventory distribution between each frequency band, filter out frequency bands with abnormal inventory changes.

[0045] S2: Obtain commodity inventory information for all oil and gas categories within the abnormal inventory change frequency band, calculate the mutual information of each pair of oil and gas categories in the current frequency band, and filter out the associated oil and gas category pairs that exceed the mutual information threshold.

[0046] S3: Calculate the current proportion of each associated oil and gas category in the total inventory, compare it with the proportion of the previous frequency band, mark the inventory change of each oil and gas category, and obtain inventory proportion difference marker data.

[0047] S4: Analyze the fluctuation pattern of inventory changes in each abnormal frequency band of inventory change, extract the fluctuation features that exceed the fluctuation threshold, and integrate them into inventory fluctuation feature data;

[0048] S5: Obtain the oil and gas categories corresponding to the inventory ratio difference marker data and inventory fluctuation characteristic data, analyze the continuity of the inventory change direction and the duration of fluctuation of the target oil and gas category, establish an adjustment ranking, and generate inventory optimization adjustment suggestions;

[0049] The abnormal inventory change frequency bands include the frequency band number, the associated oil and gas category identifier, and the range of change. The associated oil and gas category pairs include the oil and gas category pair number, the strength of the synchronization relationship, and the statistical time period. The inventory ratio difference marker data includes the direction of inventory ratio change, the magnitude of the ratio change, and the reference value of the total ratio of the combination. The inventory fluctuation characteristic data includes the frequency characteristics, the duration range, and the fluctuation component number. The inventory optimization and adjustment suggestions include the adjustment direction, the sorting priority, and the corresponding oil and gas category.

[0050] Please see Figure 2 Step S1 is as follows:

[0051] S101: Collect the inventory sequence of each oil and gas category within the monitoring period under the big data environment, obtain the original inventory records of all oil and gas categories in the continuous time dimension, and obtain the original inventory sequence of oil and gas categories.

[0052] To collect inventory sequences for each oil and gas category within a monitoring period under a big data environment, the first step is to deploy oil and gas inventory monitoring equipment and systems. This involves connecting to real-time sensor data from various oil and gas tank farms via an IoT platform, synchronizing the data to a database platform in real time, and classifying and identifying oil and gas categories such as gasoline, diesel, and liquefied natural gas. This is done by categorizing the data based on oil and gas type codes in the data fields. Then, inventory data from historical monitoring periods is retrieved, and inventory values ​​are extracted at hourly or daily granularity to ensure data integrity for the same oil and gas category within continuous periods. The original inventory fields from daily reports are extracted, and daily time tags are concatenated to generate a daily-sorted gasoline inventory time series. The same method is then used to process data sequences for other oil and gas categories, generating a time-continuous inventory record set covering all oil and gas categories.

[0053] S102: Divide the original inventory sequence of each oil and gas category into fixed time windows, extract the inventory value of each oil and gas category in each time window, and generate the time frequency band division result of oil and gas category;

[0054] To divide the original inventory sequence of each oil and gas product category into fixed time windows, the length of the time window must first be set. For example, the daily inventory data can be segmented into groups of 3 days, that is, the inventory data can be divided into different time frequency segments in chronological order: June 1 to June 3, June 4 to June 6, and June 7. For each frequency segment, the inventory values ​​of the start and end days are extracted as the representative values ​​of that frequency segment. The same operation is performed for each oil and gas product category. For example, if the inventory of diesel is value A on June 1, value B on June 2, and value C on June 3, the inventory value of this frequency segment can be extracted by calculating the average inventory value, that is, (value A + value B + value C) ÷ 3, to obtain the average inventory value of the time frequency segment. Similarly, all time periods are divided and the average inventory value or the last inventory value of each segment is calculated to generate the time frequency segment inventory sequence corresponding to diesel.

[0055] S103: Based on the time frequency band division results of oil and gas categories, calculate the difference in inventory value distribution between adjacent time frequency bands for each oil and gas category, compare it with the set inventory change threshold, identify the time frequency bands where the inventory change exceeds the threshold, and obtain the abnormal inventory change frequency bands.

[0056] Based on the time-frequency segmentation results of oil and gas categories, the difference in inventory value distribution between adjacent time-frequency segments for each oil and gas category is calculated. First, the adjacent difference in inventory value for each time period of each oil and gas category is calculated. For example, if the average inventory value of gasoline in the first time period is value D and the average inventory value in the second time period is value E, then the inventory difference is value E - value D. The sign of the difference is used to determine whether the inventory is increasing or decreasing. This inventory change value is compared with the set inventory change threshold. The inventory change threshold should be set in combination with the historical fluctuation level. For example, the absolute average of the daily inventory difference of gasoline over the past three months is taken as the benchmark difference, and then multiplied by a coefficient to set the threshold. If the absolute difference is greater than the threshold, it is identified as an abnormal frequency segment. If each time period is divided into 3 days, then every 3 days is a comparison unit. During the comparison process, it is necessary to ensure that the calculation method of the inventory value for each time period is consistent to avoid deviations in the judgment of anomalies due to different value taking methods. Finally, the abnormal frequency segments of inventory change are obtained.

[0057] Please see Figure 3 Step S2 is as follows:

[0058] S201: Obtain each time band marked in the abnormal inventory change frequency band and the corresponding oil and gas category, extract the commodity inventory information of all oil and gas categories in the current frequency band, and construct the oil and gas category inventory sequence in the abnormal inventory frequency band.

[0059] To obtain each time segment marked in the abnormal inventory change frequency range and its corresponding oil and gas category, it is first necessary to extract all time periods identified as abnormal inventory changes from the inventory change analysis results of the previous stage, and record the names of the oil and gas categories where the anomalies occurred. For example, if the inventory changes of gasoline and diesel exceed the set threshold between June 4th and June 6th, 2025, then that time period and its corresponding oil and gas category will be recorded. Subsequently, the inventory data of all oil and gas categories in these time segments are extracted, including not only the abnormal categories but also all other categories. For example, during the abnormal period, liquefied natural gas... Although no abnormal changes were observed in the gas, its inventory data was still extracted for subsequent correlation analysis. The daily inventory values ​​of these oil and gas categories within each abnormal time period were time-aligned to form a set of inventory time series data corresponding to each time period. For example, the daily inventory values ​​of gasoline from June 4th to 6th were values ​​F1, F2, and F3, which constituted the inventory series of gasoline during that time period. Similarly, the time-continuous inventory value series of categories such as diesel and natural gas were extracted. The sets of inventory values ​​of all oil and gas categories within each abnormal inventory frequency period were summarized to construct the inventory series of oil and gas categories within the abnormal inventory frequency period.

[0060] S202: Based on the inventory sequence of oil and gas categories in the abnormal inventory frequency band, construct oil and gas category pairs according to the pairwise combination method in the current frequency band, extract the inventory sequence of each oil and gas category pair in the corresponding frequency band, and calculate the mutual information value of oil and gas category pairs.

[0061] Within the abnormal inventory frequency range, inventory time series for each oil and gas category have been constructed. To analyze the correlation of inventory changes among different oil and gas categories, the inventory data needs to be paired to construct oil and gas category pairs, and then the statistical correlation index of their inventory changes—the mutual information value—needs to be calculated. Mutual information can be used to measure the degree of information sharing between two variables, and its calculation depends on the joint probability distribution and marginal probability distribution of the category inventory data. For the current inventory data scenario, the specific formula for calculating mutual information is as follows:

[0062] ;

[0063] in, : Indicates oil and gas categories Oil and gas categories The mutual information value of inventory changes within a certain inventory anomaly frequency range is a non-negative real number used to reflect the statistical dependence of inventory changes between two categories within that time period. The larger the value, the stronger the correlation. : Indicates the category of oil and gas All discrete values ​​of inventory in the inventory sequence Summation, inventory discrete values It is an oil and gas category A specific value after binning (i.e., discretizing the inventory value) within this time period, such as the range of 90 to 100 units, can be grouped into one bin. Value range; Similarly, for oil and gas categories Discrete values ​​of inventory in the inventory sequence Perform a summation of each item, each It represents the discretized inventory value of oil and gas products such as diesel and liquefied natural gas; : is the joint probability, representing the oil and gas category Inventory value And oil and gas categories Inventory value The probability of simultaneous occurrence is calculated by analyzing daily inventory levels within the statistical frequency band. The frequency of the sample is divided by the total number of samples to obtain the probability density of the joint distribution; : is the marginal probability, representing the category Inventory value The probability, without considering product category The value comes from the category. The distribution frequency of inventory values ​​within this frequency band is statistically analyzed and normalized; : is the marginal probability, representing the category Inventory value The probability, the calculation method and The same, but the statistical object is the product category. ; The ratio of the product of the joint probability and the marginal probability is represented by a logarithmic function. and The larger the deviation between the actual probability of their simultaneous occurrence and their expected probability under independent circumstances, the more significant the deviation. and The probability of both occurring simultaneously is significantly greater than the probability of them occurring independently, indicating a stronger dependency.

[0064] Let gasoline (category) ) and diesel (category) The mutual information calculation is performed step by step on the actual inventory sequence within an abnormal inventory frequency range (e.g., June 4-6, 2025, a total of 3 days).

[0065] Raw inventory data (3 days): Gasoline inventory (category) ): 100, 110, 120; Diesel inventory (category) ): 200, 210, 220; each day constitutes a joint inventory observation point, totaling 3 groups: .

[0066] By setting each inventory value to be binned in increments of 10, the following groupings are obtained: , , , , , Therefore, the joint range is: .

[0067] Calculate joint probability Because each of the three sample points appears once, the total number of samples is... ,but , , .

[0068] Calculate marginal probability and :

[0069] Gasoline inventory values ​​appear only once. , , ;

[0070] Each diesel inventory value appeared only once. , , .

[0071] The formula for mutual information is:

[0072] ;

[0073] Calculate each item in turn (because of the unpaired items) Since all three terms are 0, only the three diagonal terms are calculated.

[0074] Item 1 (Gasoline = 100, Diesel = 200): , , , , ;

[0075] Item 2 (Gasoline = 110, Diesel = 210): Same as above, also approximately ;

[0076] Item 3 (Gasoline = 120, Diesel = 220): Same as above, also approximately .

[0077] Total mutual information value .

[0078] Within this abnormal inventory frequency band, the mutual information value between gasoline and diesel inventories is: This indicates a strong correlation between them in terms of inventory changes (because each joint probability corresponds to more than three times the marginal independent probability).

[0079] During the calculation process, if a certain inventory status It has never appeared in historical data, i.e., its marginal probability. Then the inventory status With any other state The joint probability of simultaneous occurrence It must also be 0. In this case, the item... It does not contribute to mutual information and can be directly ignored in the calculation. This is because an event that has never occurred carries no information and cannot be correlated with other events. Therefore, the summation range of the above formula is actually only for all events. and The combination of inventory status.

[0080] First, oil and gas product pairs are constructed, and their inventory sequences within the abnormal frequency range are extracted. These pairs are then paired daily to form a joint inventory sample. Next, the inventory values ​​are discretized, and the frequency of each inventory value combination is counted to calculate the probability of their joint occurrence. Simultaneously, the frequency of individual inventory values ​​for each product category is counted to obtain marginal probabilities. Then, by comparing the deviation between the product of the joint probability and the marginal probability, a logarithmic function is used to quantify this deviation, forming the components of the mutual information value. All combination terms are then summed to obtain a final total mutual information value. This result reflects the correlation between the inventory changes of the two oil and gas products within the current abnormal inventory frequency range; a higher value indicates a stronger synchronicity and dependence in their inventory change trends.

[0081] S203: For each pair of oil and gas categories, compare the mutual information value with the mutual information threshold in the current frequency band, filter out oil and gas category pairs with mutual information values ​​greater than the mutual information threshold, record them as related combinations, and obtain related oil and gas category pair combinations.

[0082] The mutual information value of each pair of oil and gas commodities in the current frequency band is compared with the mutual information threshold. First, the mutual information value of each pair of oil and gas commodities needs to be obtained. This value has been calculated using the aforementioned mutual information formula. For example, the mutual information value between gasoline and diesel is the calculated result. This result is compared with the pre-set mutual information threshold. The mutual information threshold needs to be set based on historical sample data or empirical rules. For example, the average mutual information value of all oil and gas commodity pairs in the past six months in non-abnormal frequency bands can be selected and multiplied by a coefficient factor as the threshold benchmark. If the average value is M, then the mutual information threshold is M multiplied by 1.5. If the currently calculated mutual information value is greater than the threshold, it is considered that there is a significant inventory change correlation between the two commodities in this time frequency band, and the oil and gas commodity combination is recorded as a related combination.

[0083] Please see Figure 4 Step S3 is as follows:

[0084] S301: Obtain the associated oil and gas category pairs, extract the inventory values ​​of all associated oil and gas categories in the current frequency band, and calculate the proportion of the total inventory of each oil and gas category in the total inventory in the current frequency band, generating the inventory proportion data of oil and gas categories in the current frequency band.

[0085] After obtaining the associated oil and gas product category pairs, it is necessary to further extract the inventory data of all oil and gas product categories identified as related within the current frequency band. First, the individual oil and gas product categories involved in each product category pair are deduplicated and integrated to obtain a set of associated oil and gas product categories. Then, the daily inventory value of each oil and gas product category within the frequency band is calculated, and the total inventory is obtained by adding the inventory values ​​of all days. For example, the total inventory of gasoline in this frequency band is the result of adding the inventory of multiple days. The total inventory of other categories is calculated in the same way. Then, the total inventory of each oil and gas product category is divided by the sum of the total inventory of all associated oil and gas product categories within the frequency band. The formula is: Total inventory of a certain oil and gas product category ÷ Sum of the total inventory of all oil and gas product categories = Percentage of a certain oil and gas product category. The inventory percentage of each oil and gas product category within the frequency band is obtained in turn, forming the inventory percentage data of oil and gas product categories in the current frequency band.

[0086] S302: Based on the current frequency band inventory ratio data of oil and gas products, extract the total inventory of the corresponding oil and gas products in the previous frequency band, calculate the inventory ratio of the previous frequency band, call the ratio data of the current frequency band and the previous frequency band to compare the difference, and obtain the change range data of the inventory ratio of oil and gas products.

[0087] Based on the current frequency band inventory ratio data, for each oil and gas category, it is necessary to trace back its total inventory in the previous frequency band. Using the same inventory summation method as the current frequency band, the total inventory value of each oil and gas category in the previous frequency band is obtained. Then, the inventory ratio of the previous frequency band is calculated using the formula: Total inventory of oil and gas category in the previous frequency band ÷ Sum of total inventory of all oil and gas categories in the previous frequency band = Inventory ratio of the previous frequency band. Then, the inventory ratio of the current frequency band is subtracted from the inventory ratio of the previous frequency band to calculate the change in inventory ratio. The formula is: Current frequency band inventory ratio - Previous frequency band inventory ratio = Change in ratio, thus obtaining the change in inventory ratio data for each oil and gas category.

[0088] S303: Based on the data on the change in the inventory ratio of oil and gas categories, mark the direction and magnitude of inventory change for each oil and gas category, record the positive or negative value of the difference and retain the inventory value, identify the proportion difference of each oil and gas category during the frequency band change process, and obtain inventory ratio difference marking data.

[0089] Based on the data on the changes in the inventory ratio of oil and gas products, directional judgment and numerical labeling are performed for each oil and gas product category. First, the positive or negative value of the change is compared to determine whether the inventory ratio is increasing or decreasing. If the change is positive, it is marked as increasing; if it is negative, it is marked as decreasing. Then, the absolute value of the change is recorded as the inventory change rate. At the same time, the actual inventory values ​​of each oil and gas product category in the current frequency band and the previous frequency band are retained for subsequent backtracking and display. This is used to identify the difference in the inventory ratio of oil and gas products in two adjacent time frequency bands. Finally, a data set of inventory ratio difference labels containing the direction of change, the change rate, and the original inventory data is obtained.

[0090] Please see Figure 5 Step S4 is as follows:

[0091] S401: Obtain all time frequency bands and corresponding oil and gas categories marked in the abnormal inventory change frequency bands, extract the inventory time series of each oil and gas category in each abnormal frequency band, and generate the abnormal frequency band inventory sequence of oil and gas categories.

[0092] To obtain all time segments and corresponding oil and gas categories identified in the abnormal inventory change frequency bands, it is first necessary to retrieve the previously identified list of abnormal inventory frequency band markers, including the time range and oil and gas category identifier corresponding to each abnormal inventory fluctuation. The data is then extracted by frequency band. For example, if a certain category is identified as an abnormal frequency band within several consecutive or non-consecutive dates, a time-continuous inventory value sequence can be formed by calling the daily inventory records. The original inventory values ​​of that category in all abnormal frequency bands are extracted sequentially and arranged in ascending order by date to form multiple inventory time subsequences. Then, each time subsequence is aggregated by oil and gas category to generate a set of abnormal inventory time series for that oil and gas category in all abnormal time periods. Each subsequence is the inventory change trajectory of that oil and gas category in a specific abnormal frequency band.

[0093] S402: Based on the abnormal frequency band inventory sequence of oil and gas products, the empirical mode decomposition algorithm is used to perform sequence-by-sequence decomposition on each inventory sequence, extract all intrinsic mode function components, and calculate the fluctuation frequency and time distribution of each component to obtain the fluctuation frequency of oil and gas product inventory components.

[0094] The intrinsic mode function components are multiple independent fluctuation components decomposed from the inventory time series of oil and gas products in abnormal frequency bands. Each component represents the inventory change process with corresponding frequency and amplitude level.

[0095] Based on the inventory time series of oil and gas products in abnormal frequency bands, each series needs to be decomposed independently. An empirical mode decomposition (EMD) algorithm is used to decompose the inventory time series within the abnormal frequency bands into multiple intrinsic mode function (IMF) components. These components are relatively independent fluctuation units automatically generated based on local extrema within the inventory data. Each component reflects the inventory fluctuation process at a specific time scale. The fluctuation process specifically refers to the periodic upward and downward trend of the inventory value of a certain oil and gas product within a specific abnormal frequency band over time. This trend can manifest as short-term sharp changes or long-term slow fluctuations. For example, if the daily outflow of gasoline is significantly higher than the inflow in a certain frequency band, leading to a continuous decline in inventory, followed by a rebound due to replenishment and distribution, this continuous fluctuation constitutes a fluctuation process. EMF decomposition can break down the entire original inventory curve into several components with different frequencies and amplitudes. Each component represents a fluctuation level, including high-frequency, short-cycle rapid inflow and outflow changes, or low-frequency, long-cycle slow inventory accumulation and consumption trends. For each component, its fluctuation frequency within the frequency band needs to be calculated using the formula: Number of inventory changes ÷ Duration = Fluctuation Frequency. The number of inventory changes can be obtained by identifying the number of "rise-fall" or "fall-rise" repetitive segments in the sequence, and the duration is the length of the time interval corresponding to that component. Simultaneously, the distribution position of the component on the time axis needs to be extracted to determine whether the inventory fluctuation occurred before, during, or after the abnormal frequency band. The local maximum and minimum inventory values ​​corresponding to the component are also recorded to identify whether the inventory fluctuation is rapid consumption or replenishment-type. Ultimately, a multi-level fluctuation pattern is formed for each oil and gas category within the abnormal frequency band.

[0096] S403: Based on the fluctuation frequency of oil and gas product inventory components, compare the fluctuation frequency value of each component with the set fluctuation frequency threshold item by item, mark the components whose frequency value exceeds the fluctuation frequency threshold, extract the corresponding continuous segment and change pattern characteristics, and obtain inventory fluctuation characteristic data.

[0097] Based on the fluctuation frequency of each component of oil and gas inventory changes, each component needs to be screened individually, and a fluctuation frequency threshold needs to be set as a comparison benchmark. The fluctuation frequency threshold consists of two parts: the historical median frequency and the multiplier factor. The process for obtaining the historical median frequency is as follows: select multiple non-abnormal frequency range inventory sequences of the corresponding oil and gas category within a set period in the past; first, perform empirical mode decomposition on each inventory sequence; then, calculate the fluctuation frequency for each component; after summarizing all frequency values, sort them by value and select the value at the median position as the historical median frequency; the multiplier factor is usually adjusted according to the actual inventory management scenario. It is recommended to initially set it to a fixed value, such as 1.2, and then calibrate it manually or through experience. Data backtesting optimization can be achieved using the formula: Historical median frequency × multiplier factor = fluctuation frequency threshold. The frequency value of each inventory component is compared with this threshold using the formula: Component frequency - fluctuation frequency threshold = difference. If the difference > 0, it is marked as a high-frequency component. Then, the start and end times of the high-frequency component within the abnormal frequency band are recorded, and the corresponding inventory value change path within this segment is extracted. The intensity of the interval of violent inventory fluctuation is calculated using the formula: maximum inventory value - minimum inventory value = fluctuation amplitude. Combined with the time sequence, it is determined whether the overall trend of the inventory value is rising or falling. Finally, an inventory fluctuation feature dataset containing fluctuation frequency, fluctuation amplitude, time location, and trend direction is generated.

[0098] Please see Figure 6 Step S5 is as follows:

[0099] S501: Obtain the corresponding oil and gas categories from the inventory ratio difference marker data and inventory fluctuation characteristic data, extract the inventory change trend segments of each oil and gas category in the current frequency band, identify the continuity status of the inventory change direction based on the continuous rise and continuous fall direction change intervals in the inventory sequence, and generate the inventory direction continuity information of oil and gas categories.

[0100] After obtaining the corresponding oil and gas categories from the inventory ratio difference marker data and inventory fluctuation characteristic data, trend analysis needs to be performed on the inventory time series of these oil and gas categories within the current abnormal frequency band to determine whether their inventory values ​​show a significant and continuous upward or downward change. Specifically, the time series is sorted by time, and the difference between adjacent inventory values ​​is calculated point by point using the formula: Current inventory value - Previous inventory value = Inventory difference. When multiple consecutive differences are positive, it indicates a continuous upward trend in inventory; when multiple consecutive differences are negative, it indicates a continuous downward trend in inventory. After identifying these continuous segments changing in the same direction, the start and end times of each trend need to be recorded, and its duration calculated using the formula: Trend end time - Trend start time + 1 = Continuous duration. The result of each trend segment indicates whether the oil and gas category exhibits significant directional behavior within that frequency band. For example, a long period of continuous increase indicates a continuous increase in inventory, and vice versa. The final output is the inventory directional continuity information for each oil and gas category, including the direction of change, duration, and time interval.

[0101] S502: Based on the inventory direction continuity information of oil and gas categories, extract the continuous segment of the corresponding oil and gas category in the inventory fluctuation characteristics, record the time span of the component corresponding to the frequency threshold, and pair it with the direction continuity interval to determine the degree of overlap between the change trend and the fluctuation behavior, and obtain the inventory trend continuity information of oil and gas categories.

[0102] After obtaining the inventory direction continuity information for oil and gas products, it is necessary to match it with the inventory fluctuation characteristic data obtained through empirical mode decomposition mentioned earlier. The core of the matching lies in aligning the trend segments with the fluctuation components marked by the frequency threshold in the time dimension, and determining the degree of overlap between the two time intervals. The calculation method is to first extract the time range of the trend segment and the time range of the fluctuation component, determine the length of their time intersection, and use the formula: (Number of overlapping days of trend segment and component ÷ Total number of days in trend segment) × 100% = Trend overlap ratio. At the same time, the duration of the fluctuation component itself is recorded, using the formula: Fluctuation component end time - Fluctuation component start time + 1 = Fluctuation duration. The calculated trend overlap ratio indicates whether the inventory trend and inventory fluctuation behavior occur synchronously in time. The higher the ratio, the more consistent the inventory change trend and fluctuation process are, further enhancing the reliability of the trend continuity judgment. Finally, inventory trend continuity information is generated.

[0103] S503: Based on the inventory trend persistence information of oil and gas categories, the duration of the inventory change direction and the duration of the fluctuation frequency component are used as sorting criteria. The sorting priority is set, and the products are sorted according to the strength of continuity and the duration of fluctuation. The adjustment order and direction of each oil and gas category are marked, and inventory optimization adjustment suggestions are obtained.

[0104] To obtain inventory optimization and adjustment suggestions, it is necessary to comprehensively determine the adjustment priority of each oil and gas category based on the persistence of its inventory change trend and the duration of its fluctuation frequency component within the current abnormal inventory frequency range. The more stable and persistent the inventory change trend, and the more significant and persistent the fluctuation signal, the clearer the inventory behavior of that category, and the more worthy of priority intervention. Therefore, the following priority formula is constructed, which standardizes and weights these two dimensions as follows:

[0105] ;

[0106] in, : indicates the first The priority of inventory optimization and adjustment for each oil and gas category is indicated by the value; the larger the value, the higher the priority that category should be adjusted. : indicates the first The duration of continuous maintenance of the inventory trend direction for each oil and gas category (in days), that is, the time span of continuous increase or decrease in inventory. The longest continuous time value with the longest trend direction among all oil and gas categories is used for normalization to ensure that the scores of all categories are on the same scale. : indicates the first The duration (in days) of the high-frequency inventory fluctuation component of each oil and gas category, that is, the length of time that is determined to be a high-frequency fluctuation after decomposition. The maximum duration of fluctuation components across all oil and gas categories, used for normalization. Trend continuity weighting coefficient, indicating the importance of trend time in the scoring; The volatility persistence weighting coefficient represents the degree of influence of volatility behavior on the score, and must meet the following requirements: Regarding the basis for weight setting: If the system focuses more on the stability of inventory change trends, then... Set it to a larger value, for example, 0.7. A value of 0.3 is suitable for scenarios primarily focused on inventory structure optimization; however, if the focus is on the role of inventory fluctuation signals in indicating potential risks, the value should be increased. The value, for example, is set to 0.6. A value of 0.4 is suitable for dynamic portfolio rebalancing and early warning scenarios. Alternatively, the weight can be set after verifying the impact of historical adjustment results on the optimization effect.

[0107] This scoring formula normalizes and merges two inventory behavior indicators with a unified unit (day). A higher final score indicates more significant and continuous inventory behavior, which should be prioritized for inclusion in the adjustment queue. After sorting the scores in descending order, an adjustment suggestion list can be generated. Combined with the trend direction (an upward trend can be regarded as a replenishment suggestion, and a downward trend as a reduction suggestion), the inventory optimization adjustment order and operation direction for oil and gas categories can be generated.

[0108] Given: Gasoline: Inventory increased for 6 consecutive days, with high-frequency fluctuations lasting for 5 days; Diesel: Inventory decreased for 3 consecutive days, with high-frequency fluctuations lasting for 7 days; Liquefied Natural Gas: Inventory increased for 5 consecutive days, with high-frequency fluctuations lasting for 4 days. Find the longest trend and longest fluctuation periods for each of the three categories: Longest Trend Time (From gasoline), maximum fluctuation time (Based on diesel), the stability of inventory change trends is the primary factor, therefore a weight is assigned to the continuity of trends. Persistence weight of volatility Satisfying the constraint relationship .

[0109] Calculate priority one by one:

[0110] Gasoline: Trend Normalization Value = Fluctuation normalization value = ;

[0111] Priority score: ;

[0112] Diesel: Trend Normalization Value = Fluctuation normalization value = ;

[0113] Priority score: ;

[0114] Liquefied Natural Gas: Trend Normalization = Fluctuation normalization value = ;

[0115] Priority score: .

[0116] The final priority scores for the three oil and gas categories are as follows: gasoline ≈ 0.8856, liquefied natural gas ≈ 0.7282, and diesel = 0.7. The ranking results indicate that gasoline exhibits the most significant upward trend in inventory, accompanied by strong volatility signals, thus having the highest priority and being the most suitable for priority adjustment; liquefied natural gas, although exhibiting weaker volatility, has a stable trend, ranking second; and diesel, despite its volatile fluctuations, has a short-term trend, making it the lowest priority.

[0117] Within the current abnormal inventory frequency range, if the trend duration of all inventory items is 0 (i.e., no category forms a continuous upward or downward trend segment), then according to the definition of the normalization term in the original formula, the following will occur: The normalized ratio of the trend term (denominator is 0) This is not the case; similarly, if the duration of the high-frequency fluctuation components for all categories is 0, then the following will occur: To ensure that the sorting calculation remains deterministic and consistent even in boundary cases:

[0118] only The situation (no trend across all categories);

[0119] Within this abnormal frequency band, there is no continuous upward / downward trend (or the trend identification threshold is not reached).

[0120] Treat the contribution of the trend term as 0, that is, take it directly. ;

[0121] (The trend item does not contribute to the score as a whole, which is equivalent to "only the fluctuation item is determined" in this round of scoring).

[0122] only The situation (no high-frequency fluctuations across all product categories);

[0123] Within this abnormal frequency band, none of the high-frequency components marked by empirical mode decomposition reached the threshold (no results were found in frequency threshold screening in S4).

[0124] Treating the contribution of the fluctuation term as 0, i.e., directly taking... ;

[0125] (The volatility item does not contribute to the overall score; it is determined by the "trend persistence" dimension.)

[0126] and The situation (no trend, no high-frequency fluctuations);

[0127] Within this abnormal frequency band, all categories exhibit neither a directional continuous trend nor a significant duration of high-frequency fluctuations; at this point, both normalized denominators are simultaneously zero.

[0128] Define all scores for this round as 0 ( For all The overall situation is considered "stable and insensitive" and will not be included in this round of priority adjustments.

[0129] For each oil and gas category, the duration of its inventory change trend and the duration of its fluctuation frequency component within the abnormal inventory frequency range are extracted. These two parameters represent whether the inventory maintains a stable and continuous change, and whether there are continuous high-frequency disturbances, respectively. The two time values ​​are then normalized to ensure they can be weighted and calculated under the same dimension, forming a comprehensive score for ranking. By assigning weights to trend persistence and fluctuation persistence, and considering the management priorities in different scenarios, the emphasis can be flexibly shifted between trend-driven and fluctuation-driven approaches. A higher score indicates a clearer understanding of inventory changes and a greater urgency for intervention. Ranking each oil and gas category from highest to lowest score clearly identifies which categories should be prioritized in inventory adjustment plans, avoiding dispersed resource allocation or delayed intervention. Using the duration of inventory change direction and the duration of fluctuation frequency component as ranking criteria is significant because it simultaneously considers the stability and dynamic sensitivity of inventory behavior, identifying key categories with stable trends and significant disturbances, thereby improving the targeting and effectiveness of inventory management strategies.

[0130] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for dynamic optimization of oil and gas inventory based on big data, characterized in that, Comprise the following steps: S1: Collecting the inventory sequence of each oil and gas category in the monitoring period under the big data environment and dividing multiple frequency bands according to the time window, screening the inventory change abnormal frequency band according to the inventory distribution difference between each frequency band; S2: Obtain the commodity inventory information of all oil and gas categories in the inventory change abnormal frequency band, calculate the mutual information of each pair of oil and gas categories in the current frequency band, and screen the associated oil and gas category pair combination exceeding the mutual information threshold; S3: Statistic each of the associated oil and gas category pair combination in the current proportion of all inventories, compare with the proportion of the previous frequency band, mark the inventory change amplitude of each oil and gas category, and obtain the inventory proportion difference marking data; S4: Analyzing the fluctuation mode of inventory change in each abnormal frequency band in the inventory change abnormal frequency band, extracting the fluctuation characteristics exceeding the fluctuation threshold, and integrating into inventory fluctuation characteristic data; S5: Obtain the oil and gas category corresponding to the inventory proportion difference marking data and inventory fluctuation characteristic data, analyze the target oil and gas category inventory change direction continuity and fluctuation duration, and establish adjustment sorting, and generate inventory optimization adjustment suggestion; Step S5 specifically: S501: Obtain the oil and gas category corresponding to the inventory proportion difference marking data and the inventory fluctuation characteristic data, extract the inventory change trend paragraph of each oil and gas category in the current frequency band, identify the continuation state of the inventory change direction according to the direction change interval of continuous rise and continuous decline in the inventory sequence, and generate the oil and gas category inventory direction continuity information; S502: Based on the oil and gas category inventory direction continuity information, extract the continuous segment of the corresponding oil and gas category in the inventory fluctuation characteristic, record the time span of the frequency threshold corresponding component, and pair with the direction continuous interval, determine the overlap degree of change trend and fluctuation behavior, and obtain the oil and gas category inventory trend persistence information; S503: According to the oil and gas category inventory trend persistence information, the continuation time of the inventory change direction and the duration of the fluctuation frequency component are taken as the basis for sorting, the sorting priority is set, the sorting is carried out according to the strength of continuity and the duration of fluctuation, the adjustment order and adjustment direction of each oil and gas category are marked, and the inventory optimization adjustment suggestion is obtained.

2. The big data based oil and gas inventory dynamic optimization method of claim 1, wherein: The inventory change abnormal frequency band includes frequency band number, associated oil and gas category identification and change interval span, the associated oil and gas category pair combination includes oil and gas category pair number, synchronization relationship strength and statistical time period, the inventory proportion difference marking data includes inventory proportion change direction, proportion change amplitude and combination total proportion reference value, the inventory fluctuation characteristic data includes frequency characteristic, duration segment and fluctuation component number, and the inventory optimization adjustment suggestion includes adjustment direction, sorting priority and corresponding oil and gas category.

3. The big data based oil and gas inventory dynamic optimization method of claim 1, wherein: Step S1 specifically: S101: Collecting the inventory sequence of each oil and gas category in the monitoring period under the big data environment, obtaining the original inventory record of all oil and gas categories in the continuous time dimension, and obtaining the original inventory sequence of oil and gas category; S102: Divide each of the oil and gas category original inventory sequence according to a fixed time window, extract the inventory value of each oil and gas category in each time window, and generate an oil and gas category time frequency band division result; S103: Based on the oil and gas category time frequency band division result, calculate the inventory value distribution difference between adjacent time frequency bands of each oil and gas category, compare it with a set inventory change threshold, identify the time frequency bands with inventory change amplitude exceeding the threshold, and obtain inventory change abnormal frequency bands.

4. The big data based oil and gas inventory dynamic optimization method of claim 1, wherein: Step S2 is specifically: S201: Obtain each time frequency band and the corresponding oil and gas category marked in the inventory change abnormal frequency band, extract the commodity inventory information of all oil and gas categories in the current frequency band, and construct the oil and gas category inventory sequence in the inventory abnormal frequency band; S202: Based on the oil and gas category inventory sequence in the inventory abnormal frequency band, construct oil and gas category pairs in a two-by-two combination manner in the current frequency band, extract the inventory sequence of each oil and gas category pair in the corresponding frequency band, and calculate the mutual information value of the oil and gas category pair; S203: Compare the mutual information value of each pair of oil and gas categories in the current frequency band with a mutual information threshold in turn, filter the oil and gas category pairs with mutual information value greater than the mutual information threshold, record them as existing associated combinations, and obtain associated oil and gas category pair combinations.

5. The big data based oil and gas inventory dynamic optimization method of claim 4, wherein: For the mutual information value of the oil and gas category pair, the formula is: ； wherein, represents the oil and gas category with the oil and gas category represents the mutual information value of inventory change within the inventory abnormal frequency band, represents the sum of all inventory discrete values in the inventory sequence of the oil and gas category represents the sum of inventory discrete values in the inventory sequence of the oil and gas category represents the probability that the inventory value of the oil and gas category represents the probability that the inventory value of the oil and gas category represents the probability that the inventory value of the oil and gas category represents the probability that the inventory value of the oil and gas category​​​​​​​​​​​ 6. The big data based oil and gas inventory dynamic optimization method of claim 1, wherein: Step S3 is specifically: S301: Obtain the associated oil and gas category pair combination, extract the inventory value of all associated oil and gas categories in the current frequency band, and calculate the proportion of the total inventory of each oil and gas category in the current frequency band, to generate oil and gas category current frequency band inventory proportion data; S302: Based on the oil and gas category current frequency band inventory proportion data, extract the total inventory of the corresponding oil and gas category in the previous frequency band, calculate the inventory proportion of the previous frequency band, and perform difference comparison on the proportion data of the current frequency band and the previous frequency band to obtain oil and gas category inventory proportion change amplitude data; S303: According to the oil and gas category inventory proportion change amplitude data, label the inventory change direction and change amplitude of each oil and gas category, record the positive and negative of the difference value and retain the inventory value, identify the proportion difference of each oil and gas category in the frequency band change process, and obtain inventory proportion difference marking data.

7. The big data based oil and gas inventory dynamic optimization method of claim 1, wherein: Step S4 is specifically: S401: Obtain all time frequency bands and corresponding oil and gas categories marked in the inventory change abnormal frequency band, extract the inventory time sequence of each oil and gas category in each abnormal frequency band, and generate an oil and gas category abnormal frequency band inventory sequence; S402: Based on the oil and gas category abnormal frequency band inventory sequence, perform sequence-by-sequence decomposition on each inventory sequence by an empirical mode decomposition algorithm, extract all intrinsic mode function components, and calculate the fluctuation frequency and time distribution of each component respectively to obtain the oil and gas category inventory component fluctuation frequency; S403: According to the oil and gas category inventory component fluctuation frequency, compare the fluctuation frequency value of each component with a set fluctuation frequency threshold item by item, mark the components with frequency value exceeding the fluctuation frequency threshold, extract the corresponding continuous section and change pattern characteristics, and obtain inventory fluctuation characteristic data.

8. The big data based oil and gas inventory dynamic optimization method of claim 7, wherein: The intrinsic mode function components are multiple independent fluctuation components decomposed from the inventory time series of oil and gas products in abnormal frequency bands. Each component represents the inventory change process with corresponding frequency and amplitude levels.

9. The big data based oil and gas inventory dynamic optimization method of claim 1, wherein: To set sorting priority, use the following formula: ； wherein, represents the inventory optimization adjustment priority of the th oil and gas commodity, represents the continuous maintenance time of the th oil and gas commodity inventory trend direction, is the longest continuous time of the trend direction among all oil and gas commodities, represents the duration of the high-frequency inventory fluctuation component of the th oil and gas commodity, is the maximum value of the fluctuation component duration among all oil and gas commodities, represents the trend continuity weight coefficient, represents the fluctuation persistence weight coefficient.

Images