Agricultural product price prediction system based on big data

Abstract

The invention discloses an agricultural product price prediction system based on big data, and relates to the technical field of agriculture, the system constructs a price space vector field, effectively identifies an allocation mechanism between regions, calculates a driving factor, and calculates a global active index according to the allocation mechanism and the driving factor, and the module can evaluate the arbitrage opportunity between the regions in real time, so that the prediction efficiency is improved. The method comprises the steps that a global activity index is obtained, whether abnormal fluctuation exists in the market or not is judged through fluctuation of the global activity index, so that potential risks can be recognized in advance, a risk suspension instruction is triggered according to the fluctuation of the global activity index, and economic losses caused by excessive fluctuation of the market are effectively avoided; and the future transaction price of each market area can be predicted, dynamic correction can be carried out under the influence of actual market fluctuation and dispatching behaviors, and the interruption judgment module identifies the dispatching mechanism among the market areas again, so that prediction errors caused by information lag are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of agriculture, and specifically to an agricultural product price prediction system based on big data. Background Art

[0002] The prediction of agricultural product prices is an important research direction in the field of agricultural economy. Its purpose is to accurately predict the future market prices of agricultural products by using various factors such as historical data, market trends, climate change, and supply chain fluctuations. This prediction can not only help agricultural producers and suppliers optimize production and sales decisions, but also provide reliable reference data for policymakers, so as to regulate market fluctuations and improve the stability of the agricultural product market.

[0003] In the existing research on agricultural product price prediction, most traditional methods focus on price prediction in a single region. These methods mainly focus on predicting the price trend in a single market region, while ignoring the price differences and relative relationships between regions. This makes the existing models have certain limitations when predicting the price fluctuations of agricultural products in multiple regions and across regions. In order to effectively address these challenges, the price relationship between regions is particularly important. Summary of the Invention

[0004] Aiming at the deficiencies of the existing technology, the present invention provides an agricultural product price prediction system based on big data, which solves the problems in the above background art.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: An agricultural product price prediction system based on big data includes a regional collection module, a transportation identification module, a prediction correction module, and a termination determination module;

[0006] The regional collection module is used to use big data technology to obtain agricultural product transaction data in each market region during the historical stage and the real-time stage, and combine with the regional map to construct a price space vector field P;

[0007] The transportation identification module will identify the transportation mechanism between local market regions based on the price space vector field P to obtain the driving factor Adf, and calculate and obtain the global activity index Zarp according to the driving factor Adf. The fluctuation of the global activity index Zarp triggers a risk termination instruction;

[0008] The prediction correction module is used to pre-construct a prediction model to predict the predicted transaction unit price J of agricultural products in each market region in the future period, and dynamically correct the predicted transaction unit price J of agricultural products in each market region in the future period under the condition that the risk termination instruction is not triggered;

[0009] The termination determination module is used to continuously update the agricultural product transaction data in each market region, and combine with the prediction correction module to identify the transportation mechanism between local market regions again.

[0010] Preferably, the regional acquisition module includes an information acquisition unit and a construction unit;

[0011] The information acquisition unit is used to pre-select each market area to be analyzed, and through a web crawler, automatically capture the agricultural product transaction data in each market area to be analyzed during the historical stage and the real-time stage. The agricultural product transaction data includes the actual transaction price Jg of various agricultural products in each market area to be analyzed and each time point of the actual transaction price Jg; through a geographic information system, identify the adjustable paths between each market area in the regional map, and extract the adjustable path information, where the adjustable path information includes the number of segments L of each adjustable path and the length d of each segment in each adjustable path.

[0012] Preferably, the construction unit is used to identify the optimal passing path D between each market area according to the adjustable path information, specifically:

[0013]

[0014] In the formula, D ij represents the optimal passing path from the jth market area to the ith market area; L r represents the number of segments of the rth adjustable path, d q represents the length of the qth segment in the rth adjustable path, q represents the number of the road segment in the corresponding adjustable path, r represents the number of types of adjustable paths, and R j→i represents the set of all feasible paths from the jth market area to the ith market area, and both i and j represent the numbers of market areas;

[0015] By extracting the actual transaction price Jg of various agricultural products in each market area from the agricultural product transaction data, determine the spatial transaction price difference ΔJg between each market area, and combine the optimal passing path D between each market area to construct a price space vector field P. Each element in the price space vector field P represents the spatial transaction price difference ΔJg and the optimal passing path D between the corresponding market areas.

[0016] Preferably, the transportation identification module includes a path cost analysis unit, a local identification unit, and a global identification unit;

[0017] The path cost analysis unit is used to analyze whether the spatial transaction price difference ΔJg between each market area is sufficient to support the passing cost XB of the optimal passing path between the corresponding market areas according to the price space vector field P, where the passing cost XB of the optimal passing path between each market area refers to the sum of transportation cost, restriction cost, circulation processing cost, and loss cost.

[0018] Preferably, the local identification unit is used to calculate the driving factor Adf, specifically:

[0019] Adf ij = ΔJg ij -XB ij ;

[0020] Wherein, Adf ij represents the driving factor from the j-th market area to the i-th market area, and ΔJg ij represents the spatial trading price difference between the j-th market area and the i-th market area, and XB ij represents the passing cost of the optimal passing path between the j-th market area and the i-th market area;

[0021] According to the profit requirement, set a driving threshold, and compare the driving threshold with the driving factor Adf. If the driving factor Adf does not exceed the driving threshold, it indicates that there is no transfer opportunity between the corresponding market areas for the time being; if the driving factor Adf exceeds the driving threshold, it indicates that there is a transfer opportunity between the corresponding market areas, and mark the corresponding two market areas as an ordered area pair. By counting the ordered area pairs, an effective arbitrage path set X can be obtained.

[0022] Preferably, the global recognition unit is used to analyze the net arbitrage income per unit of agricultural products between each ordered area pair in the effective arbitrage path set X and the optimal passing path D between the corresponding ordered area pairs, so as to calculate and obtain the local weighted income potential value W, specifically:

[0023]

[0024] Wherein, W ij represents the local weighted income potential value from the j-th market area to the i-th market area;

[0025] Perform average normalization on the local weighted income potential values W of all ordered area pairs in the effective arbitrage path set X to calculate and obtain the global activity index Zarp, specifically:

[0026]

[0027] Wherein, X represents the effective arbitrage path set, x represents the total number of ordered area pairs that meet the arbitrage conditions in the effective arbitrage path set X, (i, j) represents the elements in the effective arbitrage path set X; W ij ' represents the local weighted income potential value from the j-th market area to the i-th market area after normalization;

[0028] Based on the local activity index Zarp in the historical period and the local activity index Zarp in the real-time stage, analyze the volatility of the global activity index Zarp to obtain the standard deviation σZarp of the global activity index, specifically:

[0029]

[0030] In the formula, T represents the total monitoring time, t represents the monitoring time point number, Zarp(t) represents the standard deviation of the global activity index at the t-th monitoring time point, and Zarp(T) avg represents the mean value of the standard deviation of the global activity index;

[0031] If the standard deviation σZarp of the global activity index exceeds three times the mean value Zarp(T) of the standard deviation of the global activity index avg it indicates that there are abnormalities in the price fluctuations of the corresponding agricultural products in each current market area. At this time, a risk suspension instruction will be triggered to stop the transportation work between ordered area pairs; if the standard deviation σZarp of the global activity index does not exceed three times the mean value Zarp(T) of the standard deviation of the global activity index avg it indicates that there are no abnormalities in the price fluctuations of the corresponding agricultural products in each current market area. At this time, the risk suspension instruction will not be triggered, indicating that the transportation work between ordered area pairs will not be stopped.

[0032] Preferably, the prediction correction module includes an initial prediction unit and a correction unit;

[0033] The initial prediction unit is used to use a linear regression model to initially construct a prediction model through the agricultural product transaction data in each market area during the historical stage. The expression of the prediction model is as follows:

[0034]

[0035] In the formula, J(z + 1) represents the predicted transaction price of each agricultural product in the corresponding market area at the future time z + 1, θ0 represents the bias term, and θ z-k+1 represents the regression coefficient of the actual transaction price of each agricultural product in the corresponding market area at the (z - k + 1)-th historical moment, Jg(z - k + 1) represents the actual transaction price of each agricultural product in the corresponding market area at the (z - k + 1)-th historical moment, M represents the total number of historical moments, k represents the historical moment number, and z represents the current time point.

[0036] Preferably, the correction unit is used to analyze whether the trend of the predicted transaction price J of the corresponding agricultural products in each market area will be behaviorally disturbed by upward or downward adjustments on the basis that the risk suspension instruction is not triggered. Specifically:

[0037] J(i, z + 1)' = J(i, z + 1) * (1 + Zy(i));

[0038] Wherein, J(i, z + 1)' represents the predicted transaction correction price of each agricultural product in the i-th market area at the future time z + 1; J(i, z + 1) represents the predicted transaction price of each agricultural product in the i-th market area at the future time z + 1, and Zy(i) represents the arbitrage scenario correction factor of each agricultural product in the i-th market area.

[0039] Preferably, the suspension determination module includes a comparison unit and a determination unit;

[0040] The comparison unit is used to continuously update the agricultural product transaction data in each market area, and extract the actual transaction price Jg corresponding to the predicted transaction correction price J(i, z + 1)' of each agricultural product in the i-th market area at the future time z + 1 in the correction unit, and subtract it from the predicted transaction correction price J(i, z + 1)' of each agricultural product in the i-th market area at the future time z + 1 in the correction unit to calculate and obtain the deviation value ΔJ.

[0041] Preferably, the determination unit is used to, when the deviation value ΔJ > 0, not trigger a risk suspension instruction, indicating that the transportation between ordered area pairs is not stopped; when the deviation value ΔJ ≤ 0, trigger a risk suspension instruction, indicating that the transportation between ordered area pairs is stopped.

[0042] The present invention provides a big data-based agricultural product price prediction system, which has the following beneficial effects:

[0043] (1) The system uses big data technology to widely obtain agricultural product transaction data in the historical and real-time stages, and combines it with regional maps. It can not only grasp the dynamic agricultural product transactions in each market area in real time, but also accurately identify the optimal travel path, which provides sufficient data support for subsequent price prediction, transportation decision-making, and market analysis, ensuring the comprehensiveness and accuracy of system prediction. The transportation identification module effectively identifies the transportation mechanism between regions by constructing the price space vector field P, calculates the driving factor Adf, and calculates the global activity index Zarp based on this. This module can evaluate the arbitrage opportunities between regions in real time and judge whether there are abnormal fluctuations in the market through the fluctuation of the global activity index Zarp. Through this mechanism, the system can identify potential risks in advance and trigger risk suspension instructions according to the fluctuation of the global activity index, effectively avoiding economic losses caused by excessive market fluctuations. The prediction correction module can not only construct a preliminary prediction model based on historical data to predict the future transaction prices of each market area, but also make dynamic corrections under the influence of actual market fluctuations and transportation behaviors. This correction process can adjust the prediction results according to real-time market feedback, ensuring the accuracy of the predicted transaction unit price and providing more targeted market regulation suggestions. After continuously updating the agricultural product transaction data of each market area, the suspension determination module, combined with the prediction correction module, re-identifies the transportation mechanism between market areas again. This continuous data update and re-identification of the transportation mechanism further enhance the flexibility and response ability of the system, enabling the system to respond to market changes in real time, optimize corresponding transportation decisions, and reduce prediction errors caused by information lag.

[0044] (2) The path cost analysis unit conducts a detailed analysis of the spatial transaction price differences between each market area according to the constructed price space vector field P to determine whether it is sufficient to support the optimal travel path cost between the corresponding market areas. This analysis function can accurately identify potential arbitrage opportunities and avoid inefficient or infeasible transportation paths. Secondly, the local identification unit quantifies the arbitrage space between each market area by calculating the driving factor Adf and promptly identifies whether there are transportation opportunities. When the driving factor Adf exceeds the preset threshold, the system marks the region as an ordered region pair, and by counting these ordered region pairs, obtains the set X of effective arbitrage paths. This process maximally improves the system's response ability and accuracy to market arbitrage behaviors, ensuring the rapid identification and implementation of arbitrage opportunities.

[0045] (3) The system first uses a linear regression model to construct a preliminary price prediction model. The model weights the historical transaction prices through regression coefficients to predict the future transaction prices of agricultural products. This process can provide reliable preliminary predicted prices for each market area based on historical data, laying the foundation for subsequent transportation decisions and market analysis. However, in the actual market, arbitrage behavior and transportation mechanisms may have a significant impact on prices. To this end, the correction unit dynamically corrects the initial prediction by introducing an arbitrage scenario correction factor. Specifically, the correction unit adjusts the predicted price based on the transportation situation between market areas, combined with the strongest local weighted profit potential value and the transfer absorption capacity index of each area. This adjustment not only takes into account historical transaction data, but also combines the real-time situation of market transportation, so that the system can more accurately reflect the changes in market supply and demand and transportation pressure.

[0046] (4) First, while the comparison unit updates the agricultural product transaction data of each market area in real time, it can compare the predicted transaction price in the prediction correction module with the actual transaction price, calculate the deviation value, and judge the effectiveness of the transportation operation based on the deviation value to avoid saturation of agricultural products in the transferred area due to excessive transportation, which ultimately causes the loss of agricultural products. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a block diagram of an agricultural product price prediction system based on big data of the present invention;

[0048] Figure 2 This is a logic diagram of an agricultural product price prediction system based on big data in the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Example 1

[0051] See also Figure 1 and Figure 2 , the present invention provides an agricultural product price prediction system based on big data, including a regional acquisition module, a transportation identification module, a prediction correction module and a suspension determination module;

[0052] The regional collection module is used to use big data technology to obtain agricultural product transaction data in each market area in the historical stage and real-time stage, identify the optimal passage between each market area in combination with the regional map, and construct the price space vector field P;

[0053] The transportation identification module will identify the transportation mechanism between local market areas based on the price space vector field P to obtain the driving factor Adf, and calculate and obtain the global activity index Zarp according to the driving factor Adf. Through the fluctuation of the global activity index Zarp, a risk suspension instruction is triggered.

[0054] The prediction correction module is used to pre-construct a prediction model to predict the predicted transaction unit price J of agricultural products in each market area in the future period, and dynamically correct the predicted transaction unit price J of agricultural products in each market area in the future period under the condition that the risk suspension instruction is not triggered.

[0055] The suspension determination module is used to continuously update the transaction data of agricultural products in each market area, and in combination with the prediction correction module, identify the transportation mechanism between local market areas again.

[0056] In this embodiment, through the regional acquisition module, the system can obtain the historical transaction data and real-time data of each market area in real time, and in combination with geographic information technology, construct the price space vector field P, so that the system can not only focus on the price trend of a single region, but also accurately capture the differences in cross-regional prices and transportation paths. In this way, the system can conduct more scientific price prediction and resource allocation according to the price differences and logistics costs between different regions, avoiding the limitations of single-market prediction in traditional methods. The introduction of the prediction correction module, combined with real-time market changes, can dynamically correct the agricultural product price prediction according to the changes in actual transaction data and arbitrage opportunities. When the market shows abnormal fluctuations, the system can respond in a timely manner and optimize the predicted value by adjusting the prediction model, thus providing a more accurate price trend prediction for market decision-makers. This real-time adjustment and correction mechanism enables the system to adapt to the rapid changes in the market and avoid the lag of traditional static prediction methods.

[0057] Combined with the price space vector field and the fluctuation of the global activity index, the transportation identification module can effectively identify potential risks in the market, and through the risk suspension instruction mechanism, timely stop ineffective or excessive transportation behaviors. Through this mechanism, the system can effectively avoid problems such as market resource waste and abnormal price fluctuations caused by excessive transportation, and ensure the stable operation of the market.

[0058] Example: Suppose in a certain market area A, the price of a certain type of agricultural product has been lower than that in the surrounding areas for a long time. Traditional prediction methods may ignore the transportation opportunities across regions, resulting in inaccurate prediction of price trends and even failing to reflect market changes in a timely manner. However, in the system of the present invention, the regional collection module first identifies the price difference between area A and the surrounding areas. Combining with the transportation identification module and the driving factor ADF, it determines that this difference can promote the situation of agricultural products being transported into area A from the surrounding areas. At the same time, it analyzes the overall arbitrage activity of the system and thereby triggers a risk suspension instruction to determine whether to stop further unreasonable transportation behaviors, thus preventing abnormal fluctuations in market prices. Through this dynamic adjustment and real-time response mechanism, the present invention can better serve the needs of market regulation and price prediction.

[0059] Embodiment 2

[0060] Please refer to Figure 1 , specifically: The regional collection module includes an information collection unit and a construction unit;

[0061] The information collection unit is used to pre-select each market area to be analyzed, and through web crawlers, automatically capture the agricultural product transaction data in each market area to be analyzed during the historical stage and the real-time stage from the Internet. The agricultural product transaction data includes the actual transaction price Jg (representing the current mainstream price level of the corresponding type of agricultural product in a certain area) of various agricultural products in each market area to be analyzed and each time point of the actual transaction price Jg; through the geographic information system, identify the adjustable paths between each market area in the regional map and extract the adjustable path information. Among them, the adjustable path information includes the number of segments L of each adjustable path and the length d of each segment in each adjustable path.

[0062] Web crawler technology can be regarded as a part of the big data technology system, but its role is more inclined to the data collection level.

[0063] The construction unit is used to identify the optimal passing path D between each market area according to the adjustable path information, specifically:

[0064]

[0065] In the formula, D ij represents the optimal passing path from the jth market area to the ith market area, that is, the shortest transportation distance; L r represents the number of segments of the rth adjustable path, d q represents the length of the qth segment in the rth adjustable path, q represents the number of the road segment in the corresponding adjustable path, r represents the number of types of adjustable paths, R j→iDenote the set of all feasible paths from the j-th market area to the i-th market area, where both i and j represent the numbers of market areas;

[0066] The optimal passage path D between each pair of market areas serves as the basis for the possible path costs in the transportation;

[0067] By extracting the actual transaction prices Jg of various agricultural products in each market area from the agricultural product transaction data, determining the spatial transaction price difference ΔJg between each pair of market areas, and combining with the optimal passage path D between each pair of market areas, construct the price space vector field P. Each element in the price space vector field P represents the spatial transaction price difference ΔJg and the optimal passage path D between the corresponding market areas.

[0068] Specifically, the transaction price difference ΔJg refers to the difference between the actual transaction prices of the corresponding agricultural products between each pair of market areas;

[0069] In this embodiment, the system automatically grabs the historical and real-time agricultural product transaction data on the Internet through web crawler technology. This process not only covers the agricultural product transaction prices in each market area but also includes detailed time series data, providing rich data support. As an important part of the big data technology system, the web crawler helps the system obtain the market transaction dynamics in real time and automatically, improving the efficiency and accuracy of data collection. Through the Geographic Information System GIS, the system can identify the allocatable paths between each pair of market areas, obtain the number of segments L of each path and the length d of each segment, and further calculate the optimal passage path from area j to area i based on this information. This path identification and optimization can help the system reasonably plan the transportation path of agricultural products, making the transportation cost of the areas where transportation can be carried out as small as possible during the transportation process, thereby improving the transportation efficiency. The system constructs the price space vector field P by identifying the transaction price difference between market areas and combining with the optimal passage path D. Each element in each vector field represents the spatial transaction price difference between market areas and the corresponding optimal passage path, providing a quantitative basis for subsequent price prediction and transportation decision-making.

[0070] Specifically, the price space vector field P provides a quantitative criterion for judging whether there is an arbitrage opportunity. After identifying the price space vector field, the system can effectively judge the transportation mechanism between each pair of market areas, thereby realizing the optimized transportation strategy based on the spatial price difference and transportation path, and further improving the allocation efficiency and cost-effectiveness of market resources.

[0071] Embodiment 3

[0072] Please refer to Figure 1 , specifically: The transportation identification module includes a path cost analysis unit, a local identification unit, and a global identification unit;

[0073] The path cost analysis unit is used to analyze whether the spatial transaction price difference ΔJg between each market area is sufficient to support the passage cost XB of the optimal passage path between the corresponding market areas according to the price space vector field P. Among them, the passage cost XB of the optimal passage path between each market area refers to the sum of transportation cost, restriction cost, circulation processing cost and loss cost.

[0074] Specifically, the transportation cost includes fuel cost and labor cost. The restriction cost refers to the toll. The circulation processing cost includes the costs of loading, unloading, packing, weighing and warehousing. The loss cost refers to the cost of the quality loss of agricultural products during the transfer. The loss cost will be calculated according to the average cost of the quality loss of the corresponding types of agricultural products during the transfer in the historical time period.

[0075] The local recognition unit is used to calculate the driving factor Adf, specifically:

[0076] Adf ij = ΔJg ij - XB ij ;

[0077] In the formula, Adf ij represents the driving factor from the j-th market area to the i-th market area. ΔJg ij represents the spatial transaction price difference between the j-th market area and the i-th market area. XB ij represents the passage cost of the optimal passage path between the j-th market area and the i-th market area;

[0078] The driving factor Adf reflects the pure profit of unit agricultural products between each pair of ordered areas.

[0079] According to the profit demand, a driving threshold is set, and the driving threshold is compared with the driving factor Adf. If the driving factor Adf does not exceed the driving threshold, it indicates that the scheduling between the corresponding market areas is not in a strong arbitrage trend, and there is no transfer opportunity between the corresponding market areas. If the driving factor Adf exceeds the driving threshold, it indicates that the scheduling between the corresponding market areas is in a strong arbitrage trend, there is a transfer opportunity between the corresponding market areas, and the corresponding two market areas are marked as an ordered area pair. By counting the ordered area pairs, an effective arbitrage path set X is obtained.

[0080] It should be noted that there is a possibility of transfer opportunities between the aforementioned corresponding market regions to drive the circulation, allocation, and trading of agricultural products between regions to achieve profits. Generally speaking, if the price of products in one region is higher than that in another region, and the transportation cost between these two regions is not high enough to eat up all the price differences, then purchasing from the region with a lower price and transporting it to the region with a higher price for sale can earn the middle profit. This behavior is called arbitrage. The situation with such conditions is called having an arbitrage opportunity. And this kind of arbitrage behavior will gradually reduce the price of agricultural products in the region with a higher price, thereby further achieving the consumption balance of the entire market. Simply put, it is to gradually reduce the price in the region with a higher price and gradually increase the price in the region with a lower price, so as to reduce the consumption differences among people and the income differences among merchants.

[0081] In this embodiment, the system judges whether the price difference between market regions is sufficient to support the transfer behavior by considering multiple factors such as transportation cost, circulation processing cost, restriction cost, and loss cost. This module avoids the selection of inefficient transfer paths through a comprehensive analysis of the transfer cost, thereby optimizing the transfer decision. For example, if the price of agricultural products in a certain market region is higher than that in a neighboring market region, but the transportation cost is too high through a long transfer path, it is difficult to achieve the expected profit. At this time, the system will automatically identify that this path is not economically feasible and avoid unnecessary cost expenditures.

[0082] Secondly, the local recognition unit evaluates the relationship between the price difference between market regions and the transfer path cost by calculating the driving factor Adf. This calculation reflects the net profit per unit of agricultural products between market regions. If the driving factor Adf exceeds the set driving threshold, it indicates that there is a strong arbitrage opportunity between the market regions, and the transfer behavior can be initiated. For example, if the price of agricultural products in region i is higher than that in region j and the cost of the transfer path is low, then the system will identify this path as a strong arbitrage path and mark it as an effective arbitrage path to transport the agricultural products in region j to region i.

[0083] Embodiment 4

[0084] Please refer to Figure 1 , specifically: The global recognition unit is used to analyze the net arbitrage income per unit of agricultural products between each pair of ordered regions according to the driving factor Adf between each pair of ordered regions in the set X of effective arbitrage paths and the optimal passing path D between the corresponding pairs of ordered regions, so as to calculate and obtain the local weighted income potential value W. Specifically:

[0085]

[0086] In the formula, W ij represents the local weighted income potential value from the j-th market region to the i-th market region;

[0087] The locally weighted revenue potential value W refers to a relative value measurement index of the actual arbitrage revenue between any two regions, considering their geographical accessibility and transportation cost control ability on the basis of the possibility of arbitrage, and is used to reflect the economic attractiveness and transportation priority of the current path.

[0088] Perform average normalization on the locally weighted revenue potential values W of all ordered region pairs in the set X of effective arbitrage paths to calculate and obtain the global activity index Zarp, specifically:

[0089]

[0090] In the formula, X represents the set of effective arbitrage paths, x represents the total number of ordered region pairs that meet the arbitrage conditions in the set X of effective arbitrage paths, (i, j) represents an element in the set X of effective arbitrage paths; W ij ’ represents the locally weighted revenue potential value from the jth market region to the ith market region after normalization;

[0091] The global activity index Zarp is an index representing the overall arbitrage activity of the system. It reflects the cluster of arbitrage opportunities among multiple regions. Simply put, the larger the global activity index Zarp, the more prevalent the arbitrage opportunities in the entire market. The more prevalent the market arbitrage opportunities, the stronger the arbitrage atmosphere of the system, and it is a macro market behavior background quantity.

[0092] Based on the local activity index Zarp during the historical period and the local activity index Zarp at the real-time stage, analyze the volatility of the global activity index Zarp to obtain the standard deviation σZarp of the global activity index, specifically:

[0093]

[0094] In the formula, T represents the total monitoring time, t represents the monitoring time point number, Zarp(t) represents the standard deviation of the global activity index at the tth monitoring time point, and Zarp(T) avg represents the mean of the standard deviation of the global activity index;

[0095] If the standard deviation σZarp of the global activity index exceeds three times the mean Zarp(T) of the standard deviation of the global activity index avg it indicates that there are abnormalities in the price fluctuations of the corresponding agricultural products in the current market regions. At this time, a risk suspension instruction will be triggered to stop the transportation work between ordered region pairs to control the arbitrage risk of the market; if the standard deviation σZarp of the global activity index does not exceed the mean Zarp(T) of the standard deviation of the global activity index avgWhen it is three times that of [specific value], it indicates that the price fluctuations of the corresponding agricultural products in each current market area do not have abnormalities for the time being. At this time, the risk suspension instruction will not be triggered, indicating that the transportation work between ordered regions will not stop.

[0096] In this embodiment, the global recognition unit calculates the local weighted revenue potential value W of each ordered region pair, comprehensively considering the net arbitrage revenue and the optimal passage path between regions. The local weighted revenue potential value is evaluated based on the relationship between the net arbitrage revenue and the transportation path cost between each region pair. This enables the system to accurately identify potential arbitrage opportunities and risk points. By normalizing the W of all ordered region pairs in the effective arbitrage path set X, the system calculates the global activity index Zarp. This index reflects the overall arbitrage activity of the system. Specifically, the larger the global activity index, the more common the arbitrage opportunities in the market, the higher the arbitrage activity of the system, and the more the market liquidity and transportation opportunities. The system also dynamically evaluates the market stability by calculating the standard deviation based on the volatility of the global activity index. When the standard deviation of the global activity index exceeds three times its mean, it indicates abnormal market fluctuations. The system will automatically trigger the risk suspension instruction, suspend the transportation behavior of relevant regions, and prevent excessive arbitrage or market collapse. This mechanism effectively controls the market arbitrage risk and ensures the long-term stable operation of the market. Through the above mechanism, the present invention provides accurate dynamic market regulation capabilities, can real-time monitor and adjust potential risks in agricultural product price prediction, and improves the market predictability and the system's risk resistance ability.

[0097] Embodiment 5

[0098] Please refer to Figure 1 , specifically: The prediction correction module includes an initial prediction unit and a correction unit;

[0099] The initial prediction unit is used to initially construct a prediction model by using a linear regression model and the agricultural product transaction data in each market area during the historical stage. The expression of the prediction model is as follows:

[0100]

[0101] In the formula, J(z + 1) represents the predicted transaction price of each agricultural product in the corresponding market area at the future time point z + 1. θ0 represents the bias term, reflecting the conventional price level or structural cost offset under the overall market background. θ z-k+1 represents the regression coefficient of the actual transaction price of each agricultural product in the corresponding market area at the (z - k + 1)-th historical moment, representing the influence degree of a certain historical time point on the future price. Jg(z - k + 1) represents the actual transaction price of each agricultural product in the corresponding market area at the (z - k + 1)-th historical moment. M represents the total number of historical moments. k represents the historical moment number. z represents the current time point.

[0102] The correction unit is used to analyze whether the trend of the predicted transaction price J of the corresponding agricultural products in each market area will be behaviorally disturbed by upward or downward adjustments under the influence of the current transportation work (i.e., arbitrage behavior) on the basis that the risk suspension instruction is not triggered. Specifically:

[0103] J(i, z + 1)' = J(i, z + 1) * (1 + Zy(i));

[0104] In the formula, J(i, z + 1)' represents the predicted transaction correction price of each agricultural product in the i-th market area at the future time z + 1; J(i, z + 1) represents the predicted transaction price of each agricultural product in the i-th market area at the future time z + 1, and Zy(i) represents the arbitrage scenario correction factor of each agricultural product in the i-th market area; Zy(i) may be positive (upward adjustment) or negative (downward adjustment);

[0105] Specifically, after introducing Zy(i), the prediction model has situation awareness, can automatically adapt to the behavior-driven changes in the current market, improve the response ability of the prediction to factors such as transportation and arbitrage, and expand the interpretation space of the prediction model, from pure data prediction to a joint data and behavior prediction model;

[0106] Among them, the arbitrage scenario correction factor Zy(i) of each agricultural product in the i-th market area is obtained through the following formula:

[0107] Zy(i) = log[1 + Zarp * W top * (Xn(i) - 1)];

[0108] In the formula, W top represents the strongest local weighted revenue potential value of the i-th market area, and Xn(i) represents the import absorption capacity index (market digestion capacity) of the i-th market area; log[*] represents the logarithmic function, which is used for function compression to prevent the adjustment factor from being non-linearly amplified;

[0109] Only when the three conditions of the local activity index, the strongest local weighted revenue potential value, and the import absorption capacity index are simultaneously met, it indicates that the import behavior will cause a large degree of regional price fluctuations. If any one of them is a low value, the overall result approaches 0, indicating that the arbitrage scenario has no substantial disturbance to the price prediction. And because the product result may be abnormally amplified due to one item being too large, the log function is used because the log function has the ability to suppress the amplification effect;

[0110] Dynamically correct the original prediction model through the arbitrage scenario correction factor;

[0111] The import absorption capacity index Xn(i) of the i-th market area is obtained through the following formula:

[0112]

[0113] Where Dmd(i) represents the local market demand for the corresponding agricultural products in the i-th market area. represents the total amount of corresponding agricultural products transferred from other market areas to the i-th market area, y represents the market area number transferred to the i-th market area, and V represents the total number of market areas transferred to the i-th market area;

[0114] The absorption capacity index can reflect whether the corresponding transfer area currently has sufficient market "absorption capacity";

[0115] If Xn(i)>1, the amount of transfer will be far lower than the local demand, indicating that the pressure of transfer has not been released, and the price forecast will be adjusted upward. If Xn(i)≤1, then (Xn(i)-1)≤0, Zarp*W top *(Xn(i)-1)≤0, indicating that the amount of transfer has met the absorption capacity, supply exceeds demand, and the price forecast is adjusted downward, but there is a mathematical risk at this time: if Zarp*W top *(Xn(i)-1)≤1, then log[1+] will be invalid (negative logarithm), which may indicate an abnormal supply and demand relationship in the market or excessive transportation behavior, and the transportation operation will be stopped at this time.

[0116] It should be noted that for a market area, there may be multiple sources of market area transportation paths, each path has a corresponding local weighted profit potential value, so the strongest local weighted profit potential value W of the i-th market area mentioned above is top It refers to the local weighted profit potential value with the largest value corresponding to the i-th market area. In the arbitrage adjustment behavior, the strongest adjustment direction is often the real driver of price fluctuations.

[0117] In this embodiment, first, the initial prediction unit uses historical transaction data to preliminarily construct a prediction model through a linear regression model to predict the prices of agricultural products in various market areas in the future period. Through this method, the system can derive the preliminary trend of market prices with the support of a large amount of historical data, providing a basis for subsequent decision-making. However, this prediction only takes into account historical data, and does not take into account the dynamic changes in operation and supply and demand between regions. Therefore, the introduction of the correction unit is particularly important. The correction unit further analyzes whether the original prediction results need to be dynamically adjusted under the influence of arbitrage behavior. The system calculates the arbitrage scenario correction factor and makes prediction corrections based on the current market area's absorption capacity and local weighted profit potential.

[0118] Specifically, if the amount of transferred-in goods in a region is lower than the local demand (the transferred-in pressure has not been released), it indicates that the market is still in a state of supply falling short of demand, and the system will moderately raise the price prediction. If the transferred-in amount has approached or exceeded the local demand and the market is approaching saturation, the price prediction may be lowered. Through this dynamic correction mechanism, the present invention effectively solves the defect in traditional prediction methods of ignoring market transportation behavior and supply-demand changes, ensures that the system can flexibly respond to market changes, and provides more accurate price predictions. This system not only optimizes the market regulation of agricultural products but also provides reliable data support for decision-makers, helps improve the allocation efficiency of market resources, and reduces risks.

[0119] Embodiment 6

[0120] Please refer to Figure 1 , specifically: The suspension determination module includes a comparison unit and a determination unit;

[0121] The comparison unit is used to continuously update the agricultural product transaction data in each market region, and extract the actual transaction price Jg corresponding to the predicted transaction correction price J(i, z + 1)' of each agricultural product in the i-th market region at the future time z + 1 in the correction unit, and subtract it from the predicted transaction correction price J(i, z + 1)' of each agricultural product in the i-th market region at the future time z + 1 in the correction unit to calculate and obtain the deviation value ΔJ.

[0122] The determination unit is used to, when the deviation value ΔJ > 0, not trigger a risk suspension instruction, indicating that the transportation work between ordered region pairs is not stopped; when the deviation value ΔJ ≤ 0, trigger a risk suspension instruction, indicating that the transportation work between ordered region pairs is stopped.

[0123] In this embodiment, first, while the comparison unit is real-time updating the agricultural product transaction data in each market region, it can compare the predicted transaction price in the prediction correction module with the actual transaction price. By calculating the deviation value (i.e., the difference between the predicted price and the actual transaction price), the comparison unit can immediately identify the accuracy of the price prediction. When the deviation value is positive, it indicates that the actual transaction price is higher than the predicted price, indicating that the current market transportation needs to continue; conversely, if the deviation value is negative, it indicates that the actual transaction price is lower than the predicted price, indicating that the current market transportation does not need to continue. This mechanism can judge the effectiveness of the transportation operation and avoid the saturation of agricultural products in the transferred-in region due to excessive transportation, ultimately resulting in the loss of agricultural products.

[0124] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An agricultural product price prediction system based on big data, characterized in that: It includes a regional acquisition module, a transportation identification module, a prediction correction module, and a termination determination module; The regional acquisition module is used to utilize big data technology to obtain agricultural product transaction data in each market region during the historical stage and the real-time stage, and combine it with the regional map to construct a price space vector field P; The transportation identification module will identify the transportation mechanism between local market regions based on the price space vector field P to obtain the driving factor Adf, and calculate and obtain the global activity index Zarp according to the driving factor Adf. The fluctuation of the global activity index Zarp triggers a risk termination instruction; The prediction correction module is used to pre-construct a prediction model to predict the predicted transaction unit price J of agricultural products in each market region in the future period, and dynamically correct the predicted transaction unit price J of agricultural products in each market region in the future period under the condition that the risk termination instruction is not triggered; The termination determination module is used to continuously update the agricultural product transaction data in each market region, and combine it with the prediction correction module to re-identify the transportation mechanism between local market regions.

2. The agricultural product price prediction system based on big data according to claim 1, characterized in that: The regional acquisition module includes an information acquisition unit and a construction unit; The information acquisition unit is used to pre-select each market region to be analyzed, and automatically crawl the agricultural product transaction data in each market region to be analyzed during the historical stage and the real-time stage from the Internet through web crawlers. The agricultural product transaction data includes the actual transaction price Jg of various agricultural products in each market region to be analyzed and each time point of the actual transaction price Jg; through the geographic information system, identify the deployable paths between each market region in the regional map, and extract the deployable path information, where the deployable path information includes the number of segments L of each deployable path and the length d of each segment in each deployable path.

3. The agricultural product price prediction system based on big data according to claim 2, wherein: The construction unit is used to identify the optimal passing path D between each market region according to the deployable path information, specifically: Where D ij represents the optimal passing path from the j-th market area to the i-th market area; L r represents the number of segments of the r-th adjustable path, d q represents the length of the q-th segment in the r-th adjustable path, q represents the number of the road segment in the corresponding adjustable path, r represents the number of types of adjustable paths, R j→i represents the set of all feasible paths from the j-th market area to the i-th market area, and both i and j represent the numbers of market areas; By extracting the actual transaction price Jg of various agricultural products in each market region from the agricultural product transaction data, determine the spatial transaction price difference ΔJg between each market region, and combine it with the optimal passing path D between each market region to construct a price space vector field P. Each element in the price space vector field P represents the spatial transaction price difference ΔJg and the optimal passing path D between the corresponding market regions.

4. The agricultural product price prediction system based on big data according to claim 3, characterized in that: The transportation identification module includes a path cost analysis unit, a local identification unit, and a global identification unit; The path cost analysis unit is used to analyze whether the spatial transaction price difference ΔJg between each market region is sufficient to support the passing cost XB of the optimal passing path between the corresponding market regions according to the price space vector field P, where the passing cost XB of the optimal passing path between each market region refers to the sum of the transportation cost, the restriction cost, the circulation processing cost, and the loss cost.

5. The agricultural product price prediction system based on big data according to claim 4, wherein: The local identification unit is used to calculate the driving factor Adf, specifically: Adf ij = ΔJg ij -XB ij ; where Adf ij represents the driving factor from the j-th market area to the i-th market area, and ΔJg ij represents the spatial trading price difference between the j-th market area and the i-th market area, and XB ij represents the passing cost of the optimal passing path between the j-th market area and the i-th market area; According to the profit demand, the driving threshold is set and compared with the driving factor Adf. If the driving factor Adf does not exceed the driving threshold, it indicates that there is no transportation opportunity between the corresponding market areas. If the driving factor Adf exceeds the driving threshold, it indicates that there is a transportation opportunity between the corresponding market areas, and the corresponding two market areas are marked as ordered area pairs. By counting the ordered area pairs, the effective arbitrage path set X is obtained.

6. The agricultural product price prediction system based on big data according to claim 5, characterized in that: The global identification unit is used to analyze the unit agricultural product arbitrage net income between each ordered region pair according to the driving factor Adf between each ordered region pair in the effective arbitrage path set X and the optimal passage path D between the corresponding ordered region pairs, so as to calculate the local weighted income potential value W, which is: where, W ij represents the local weighted revenue potential value from the j-th market area to the i-th market area; The local weighted profit potential values ​​W of all ordered region pairs in the effective arbitrage path set X are averaged and normalized to calculate the global activity index Zarp, which is: Wherein, X represents the set of effective arbitrage paths, x represents the total number of ordered regional pairs that satisfy the arbitrage conditions in the set of effective arbitrage paths X, and (i, j) represents an element in the set of effective arbitrage paths X; W ij ’ represents the local weighted return potential value from the j-th market region to the i-th market region after normalization; Based on the bureau activity index Zarp in the historical period and the bureau activity index Zarp in the real-time stage, the volatility of the global activity index Zarp is analyzed to obtain the standard deviation σZarp of the global activity index, which is: Wherein, T represents the total monitoring time, t represents the monitoring time point number, Zarp(t) represents the standard deviation of the global activity index at the t-th monitoring time point, and Zarp(T) avg represents the mean value of the standard deviation of the global activity index; If the standard deviation σZarp of the global activity index exceeds three times the mean Zarp(T) of the standard deviation of the global activity index avg it indicates that there are abnormalities in the price fluctuations of the corresponding agricultural products in the current market regions. At this time, a risk suspension instruction will be triggered to stop the transportation work between ordered area pairs; If the standard deviation σZarp of the global activity index does not exceed three times the mean Zarp(T) of the standard deviation of the global activity index avg it indicates that the price fluctuations of the corresponding agricultural products in the current market regions are not abnormal for the time being. At this time, the risk suspension instruction will not be triggered, indicating that the transportation between the orderly regional pairs will not be stopped.

7. The agricultural product price prediction system based on big data according to claim 6, wherein: The prediction and correction module includes an initial prediction unit and a correction unit; The initial prediction unit is used to use the linear regression model to initially construct a prediction model through the agricultural product transaction data in each market area in the historical stage. The expression of the prediction model is as follows: In the formula, J(z + 1) represents the predicted transaction prices of various agricultural products in the corresponding market area at the future time z + 1, θ0 represents the bias term, and θ z-k+1 represents the regression coefficient of the actual transaction prices of various agricultural products in the corresponding market area at the (z - k + 1)-th historical moment, Jg(z - k + 1) represents the actual transaction prices of various agricultural products in the corresponding market area at the (z - k + 1)-th historical moment, M represents the total number of historical moments, k represents the historical moment number, and z represents the current time point.

8. The agricultural product price prediction system based on big data according to claim 7, characterized in that: The correction unit is used to analyze whether the trend of the predicted transaction price J of the corresponding agricultural products in each market area will be subject to behavioral disturbances of upward and downward adjustments under the influence of the current transportation work, on the basis of not triggering the risk suspension order. Specifically: J(i,z+1)'=J(i,z+1)*(1+Zy(i)); In the formula, J(i, z+1)' represents the predicted transaction price of each agricultural product in the i-th market area at the future time z+1; J(i, z+1) represents the predicted transaction price of each agricultural product in the i-th market area at the future time z+1, and Zy(i) represents the arbitrage scenario correction factor of each agricultural product in the i-th market area.

9. The agricultural product price prediction system based on big data according to claim 8, characterized in that: The suspension determination module includes a comparison unit and a determination unit; The comparison unit is used to continuously update the agricultural product transaction data in each market area, and extract the actual transaction price Jg corresponding to the predicted transaction revised price J(i, z+1)' of each agricultural product in the i-th market area at the future time z+1 in the correction unit, and subtract it from the predicted transaction revised price J(i, z+1)' of each agricultural product in the i-th market area at the future time z+1 in the correction unit to calculate the deviation value ΔJ.

10. The agricultural product price prediction system based on big data according to claim 9, characterized in that: The determination unit is used for not triggering the risk termination instruction when the deviation value ΔJ>0, indicating that the transportation work between the ordered area pairs will not be stopped; when the deviation value ΔJ≤0, triggering the risk termination instruction, indicating that the transportation work between the ordered area pairs will be stopped.

Images