Grain storage management method and system

Abstract

The present invention relates to the field of grain supervision technology, and in particular to a grain storage management method and system, the method comprising the following steps: collecting historical price data and historical demand data of grain, and collecting real-time images of grain warehouses at the same time; constructing historical price sequences and historical demand sequences, and then constructing a mathematical fluctuation model between grain prices and grain demand; predicting future grain prices based on historical price sequences, and predicting future grain demand based on historical demand sequences; using mathematical fluctuation models to verify future grain prices and future grain demand, and storing grain according to the verification results; and monitoring and early warning of grain status based on real-time images of grain warehouses based on deep learning algorithms. The present invention uses a mathematical fluctuation model to verify grain demand and price, and then stores grain according to the verification results, has strong foresight, systematicness and flexibility, and can control grain storage costs and reduce grain storage risks.

Description

Technical Field

[0001] The present invention relates to the technical field of grain supervision, and in particular to a grain storage management method and system. Background Art

[0002] Prior art usually predicts grain prices or grain demand before grain storage, and then stores grain based on the predicted results. However, this method of grain storage has some defects. First, the prior art has obvious shortcomings in grain price forecasting. Traditional forecasting methods lack systematic data analysis and scientific forecasting models, resulting in inaccurate forecasting results and difficulty in providing reliable basis for storage decisions. The uncertainty of such forecasts makes it difficult for warehouse managers to make timely and effective inventory adjustments when facing market price fluctuations, thereby increasing the risk and cost of grain storage. Secondly, the prior art also has limitations in grain demand forecasting. With the strengthening of the trend of diversification and personalization of consumer demand, traditional demand forecasting methods can no longer meet the needs of modern storage management. Grain demand may fluctuate dramatically in a short period of time, and the prior art often cannot accurately capture such changes, resulting in a disconnect between inventory management strategies and actual market demand, which in turn causes inventory backlogs or shortages, which also increases the risk and cost of grain storage. Summary of the invention

[0003] In view of the defects in the prior art, the present invention provides a grain storage management method and system.

[0004] In order to achieve the above-mentioned purpose, in the first aspect, the present invention provides a method for grain storage management, the method comprising the following steps: collecting historical price data and historical demand data of grain, and collecting real-time images of grain warehouses at the same time; constructing historical price sequences and historical demand sequences, and then constructing a mathematical fluctuation model between grain prices and grain demand; predicting future grain prices based on the historical price sequences, and predicting future grain demand based on the historical demand sequences; using the mathematical fluctuation model to verify the future grain prices and the future grain demand, and storing grain according to the verification results; and monitoring and warning the grain status based on the real-time images of the grain warehouses based on deep learning algorithms. The present invention uses a mathematical fluctuation model to verify grain demand and price, and then stores grain according to the verification results, which has strong foresight, systematicness and flexibility, and can control grain storage costs and reduce grain storage risks.

[0005] Optionally, constructing a historical price sequence and a historical demand sequence, and then constructing a mathematical fluctuation model between food prices and food demand, comprises the following steps:

[0006] Constructing a historical price sequence based on the historical price data, and constructing a historical demand sequence based on the historical demand data;

[0007] The historical price sequence and the historical demand sequence are used to construct a volatility index sequence, and then the mathematical volatility model is obtained based on a Gaussian mixture model and the Bayesian information criterion.

[0008] Furthermore, the present invention reflects the fluctuation relationship between food prices and demand by constructing a mathematical fluctuation model, thereby providing a basis for subsequent accurate food demand forecasting, ensuring the rationality and accuracy of food storage, and thus controlling food storage costs and reducing food storage risks.

[0009] Optionally, the step of constructing a volatility index sequence using the historical price sequence and the historical demand sequence, and then acquiring the mathematical volatility model based on a Gaussian mixture model and a Bayesian information criterion comprises the following steps:

[0010] The quotient of the grain price and the grain demand in the same storage period is used as a volatility index, and then the historical price series and the historical demand series are used to construct a volatility index series;

[0011] Dividing the volatility index sequence into a first subsequence, a second subsequence, a third subsequence and a fourth subsequence according to season;

[0012] The corresponding mathematical fluctuation model is obtained based on the four divided subsequences respectively and based on the Gaussian mixture model and the Bayesian information criterion.

[0013] Furthermore, the fluctuation index sequence is divided into four subsequences according to season, and then the corresponding mathematical fluctuation model of each subsequence is obtained, so that the obtained mathematical fluctuation model has seasonal characteristics, which improves the seasonal specificity of the mathematical fluctuation model and provides a basis for subsequent accurate food demand forecasting.

[0014] Optionally, predicting future food prices according to the historical price sequence and predicting future food demand according to the historical demand sequence comprises the following steps:

[0015] Constructing a food price prediction model based on the historical price series, and using the food price prediction model to predict future food prices;

[0016] A food demand forecasting model is constructed according to the historical demand sequence, and the food demand forecasting model is used to forecast future food demand.

[0017] Furthermore, the present invention provides data support for grain storage by predicting future grain prices and future grain demand.

[0018] Optionally, constructing a food price prediction model according to the historical price sequence, and using the food price prediction model to predict future food prices comprises the following steps:

[0019] Dividing the historical price sequence into a first price sequence and a second price sequence, and preprocessing the first price sequence using variational mode decomposition to obtain a first component set;

[0020] Using the first component in the first component set as an input of a recurrent neural network, and using the predicted value of the grain price as an output of the recurrent neural network, and then constructing an initial grain price prediction model based on the variational mode decomposition and the recurrent neural network;

[0021] Based on the second price sequence, the particle swarm optimization algorithm is used to optimize the core parameters of variational mode decomposition on the basis of the initial food price prediction model, so as to obtain the food price prediction model and predict the future food price.

[0022] Furthermore, the use of variational mode decomposition can decompose the historical price series into multiple IMF components with different frequency scales. These components can better reflect the volatility characteristics of the data at different time scales, making the food prices predicted using recurrent neural networks more accurate.

[0023] Optionally, constructing a food demand forecasting model according to the historical demand sequence, and using the food demand forecasting model to forecast future food demand comprises the following steps:

[0024] Calculate the average historical demand for grain for a given number of storage cycles;

[0025] Considering the impact of the period intervals between different grain historical transaction periods on the historical demand, a grain demand forecasting model is constructed in combination with the average historical demand. The grain demand forecasting model satisfies the following relationship:

[0026]

[0027] Wherein, n is the number of storage cycles, is the historical demand data of the i-th storage cycle under the set number of storage cycles, , For the future food demand, is the average historical demand.

[0028] Furthermore, the food demand forecasting model takes into account the impact of the periodic intervals between food trading cycles on food demand. Even when food demand fluctuates drastically, it can calculate a more accurate future food demand, avoid food inventory backlogs or shortages, and reduce food storage risks and costs.

[0029] Optionally, the using the mathematical fluctuation model to verify the future food price and the future food demand, and storing food according to the verification result comprises the following steps:

[0030] Calculate the quotient of the future food price and the future food demand, and record the calculation result as the volatility index forecast value;

[0031] According to the season corresponding to the fluctuation index prediction value, the fluctuation index prediction value is brought into the corresponding mathematical fluctuation model to calculate the fluctuation determination factor prediction value, and it is determined whether the fluctuation determination factor prediction value is within the fluctuation range;

[0032] If the predicted value of the fluctuation determination factor is within the fluctuation range, the grain is collected and stored in the warehouse according to the future grain demand;

[0033] If the predicted value of the fluctuation determination factor is not within the fluctuation range, the reserve grain storage strategy is adopted to store the grain in the warehouse.

[0034] Furthermore, mathematical fluctuation models are used to verify future food prices and future food demand, so that warehouse managers can more keenly capture market changes and then store food according to different food storage plans, avoiding inventory backlogs or shortages caused by the disconnection between inventory management strategies and actual market demand, and reducing food storage risks and costs.

[0035] Optionally, the reserve food storage strategy comprises the following steps:

[0036] If the future food demand is less than the current food demand, the food is stored in the warehouse according to the current food demand;

[0037] If the future food demand is greater than the current food demand, the food will be stored in the first warehouse according to the current food demand, and the portion of the future food demand that exceeds the current food demand will be stored in other warehouses that are closest to the first warehouse.

[0038] Furthermore, the use of alternative grain storage strategies can improve the flexibility of grain storage, ensure the stability of grain supply, optimize the use of warehouse resources, and improve grain storage efficiency.

[0039] Optionally, the monitoring and early warning of grain status based on the real-time image of the grain warehouse and the deep learning algorithm comprises the following steps:

[0040] A grain status monitoring dataset was constructed using grain images in different grain states, and an initial monitoring model was constructed using a convolutional neural network.

[0041] Using the grain status monitoring data set to complete the training and verification of the initial monitoring model to obtain a grain monitoring model;

[0042] The grain status is identified according to the real-time image using the grain monitoring model, and an alarm is issued when the grain status is abnormal.

[0043] Furthermore, real-time images and deep learning algorithms are used to monitor the state of grain in real time, and timely alarms are issued when the grain state is abnormal, so that relevant personnel can take timely countermeasures and reduce economic losses.

[0044] In summary, the present invention firstly constructs a mathematical fluctuation model to reflect the mathematical fluctuation model between food prices and demand, thereby providing a basis for subsequent accurate food demand forecasting, and then uses the food price forecasting model and the food demand forecasting model to predict future food prices and future food demand respectively, and then uses the mathematical fluctuation model to verify the forecasting results, so that warehouse managers can more keenly capture market changes, and then store food according to different food warehousing storage plans, so that the method has strong foresight, systematicness and flexibility, avoids the problem of inventory backlog or shortage caused by the disconnection between inventory management strategies and actual market demand, and reduces food storage risks and costs.

[0045] In a second aspect, the present invention further provides a grain storage management system, which uses a grain storage management method provided by the present invention, and the system includes: a data acquisition module, which is used to collect historical price data and historical demand data of grain, and at the same time collect real-time images of grain warehouses; a grain prediction module, which is used to construct historical price sequences and historical demand sequences, and then construct a mathematical fluctuation model between grain prices and grain demands; predict future grain prices based on the historical price sequences, and at the same time predict future grain demands based on the historical demand sequences; use the mathematical fluctuation model to verify the future grain prices and the future grain demands, and store grain according to the verification results; a monitoring and early warning module, which is used to monitor and warn the grain status based on the real-time images of the grain warehouses based on deep learning algorithms; a data storage module, which is used to store data obtained by the data acquisition module, the grain prediction module and the monitoring and early warning module; a data output module, which is used to output the real-time images, the mathematical fluctuation model, the future grain prices, the future grain demands and the grain storage plan. This system can stably execute the grain storage management method provided by the present invention, improve the efficiency of grain storage, and improve the practicality of the method of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 A schematic diagram of a grain storage management method according to an embodiment of the present invention;

[0048] Figure 2 A simplified flow chart of obtaining optimal core parameters of variational mode decomposition according to an embodiment of the present invention;

[0049] Figure 3 The figure is a schematic diagram of the framework of a grain storage management system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are only for illustration and are not intended to limit the present invention. In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it is obvious to those of ordinary skill in the art that these specific details do not need to be adopted to implement the present invention. In other examples, in order to avoid confusing the present invention, known circuits, software or methods are not specifically described.

[0051] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "in one embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. In addition, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combination and / or subcombination. In addition, it should be understood by those of ordinary skill in the art that the figures provided herein are for illustrative purposes and that the figures are not necessarily drawn to scale.

[0052] It should be noted in advance that, in an optional embodiment, except for independent explanations, the same symbols or letters appearing in all formulas have the same meanings and values.

[0053] In an alternative embodiment, see Figure 1 The present invention provides a grain storage management method, the method comprising the following steps:

[0054] S1. Collect historical price data and historical demand data of grain in different storage periods, and collect real-time images of grain warehouses.

[0055] Specifically, in this embodiment, historical price data and historical demand data of grain in different storage periods are collected on the Internet or in periodicals, and surveillance cameras are used to collect real-time images of the grain warehouse.

[0056] Furthermore, historical demand data refers to the historical demand for grain, and the grain demand mentioned later refers to the grain demand; a storage cycle is the time interval from the entry of grain into the warehouse to the exit of grain from the warehouse; the collected historical price data and historical demand data should include historical price data and historical demand data for the entire year.

[0057] S2. Construct historical price series and historical demand series, and then construct a mathematical fluctuation model between food prices and food demand.

[0058] Among them, S2 specifically includes the following steps:

[0059] S21. Construct a historical price sequence according to the historical price data, and construct a historical demand sequence according to the historical demand data.

[0060] S22. Use the historical price sequence and the historical demand sequence to construct a volatility index sequence, and then obtain the mathematical volatility model based on a Gaussian mixture model and the Bayesian information criterion.

[0061] Among them, S22 specifically includes the following steps:

[0062] S221. Taking the quotient of the grain price and the grain demand in the same storage cycle as a volatility index, and then using the historical price sequence and the historical demand sequence to construct a volatility index sequence.

[0063] Specifically, in this embodiment, the data in the historical price sequence and the historical demand sequence are one-to-one corresponding in time, so the volatility index of each storage cycle can be calculated, and then the volatility index sequence can be constructed.

[0064] S222. Divide the volatility index sequence into a first subsequence, a second subsequence, a third subsequence and a fourth subsequence according to season.

[0065] S223. Based on the four divided subsequences, obtain the corresponding mathematical fluctuation model based on the Gaussian mixture model and the Bayesian information criterion.

[0066] Specifically, in this embodiment, for ease of understanding and explanation, the first subsequence is taken as an example to construct a corresponding mathematical fluctuation model.

[0067] 1. Randomly initialize the Gaussian mixture model and parameters such as mean, variance, and number of clusters.

[0068] 2. Based on the first subsequence, the expectation-maximization algorithm is used to iteratively update the parameters of the Gaussian mixture model.

[0069] 3. Calculate the probability density function of each data point in the first subsequence according to the updated parameters, and then obtain the first Gaussian mixture model of the first subsequence.

[0070] 4. The maximum likelihood function of the Gaussian mixture model is calculated. Since the number of clusters and the sample size of the first subsequence are known, the BIC value of the Gaussian mixture model is calculated using the Bayesian Information Criterion.

[0071] 5. Select a new number of clusters and repeat steps 1-4 to obtain multiple Gaussian mixture models and corresponding BIC values. The Gaussian mixture model with the smallest BIC value is taken as the best Gaussian mixture model.

[0072] 6. Use the best Gaussian mixture model as the mathematical fluctuation model of the first subsequence.

[0073] Furthermore, this embodiment reflects the fluctuation relationship between food prices and demand by constructing a mathematical fluctuation model, thereby providing a basis for subsequent accurate food demand forecasting, ensuring the rationality and accuracy of food storage, and thereby controlling food storage costs and reducing food storage risks.

[0074] S3. Predicting future food prices based on the historical price sequence, and predicting future food demand based on the historical demand sequence.

[0075] The future grain price is the grain price in the next storage cycle, and the future grain demand is the grain demand in the next storage cycle. S3 specifically includes the following steps:

[0076] S31. Construct a food price prediction model based on the historical price sequence, and use the food price prediction model to predict future food prices.

[0077] Among them, S31 specifically includes the following steps:

[0078] S311. Divide the historical price sequence into a first price sequence and a second price sequence, and use variational mode decomposition to preprocess the first price sequence to obtain a first component set.

[0079] Specifically, in this embodiment, the sample capacity in the first price series and the second price series is the same, and the first price series can be decomposed into multiple first IMF components with different frequency scales, namely, first components, by using variational mode decomposition (VMD). The set of these first IMF components is the first component set.

[0080] S312. Use the first component in the first component set as the input of the recurrent neural network, and the predicted value of the grain price as the output of the recurrent neural network, and then construct an initial grain price prediction model based on the variational mode decomposition and the recurrent neural network.

[0081] Specifically, in this embodiment, the first IMF component is used as the input of the recurrent neural network, and the predicted value of the grain price is used as the output of the recurrent neural network. Then the first component set is divided into a training set and a verification set to complete the training and verification of the recurrent neural network. The initial grain price prediction model can be obtained by using variational mode decomposition and the trained and verified recurrent neural network, and the initial grain price prediction model is recorded as a CVMD-LSTM model.

[0082] Furthermore, the use of variational mode decomposition can decompose the first price series into multiple IMF components with different frequency scales, which can better reflect the fluctuation characteristics of the data at different time scales, so that the future food prices predicted using the recurrent neural network have higher accuracy. The specific type of recurrent neural network is LSTM, and the use of the optimal IMF component to complete the training of LSTM is a prior art, which will not be described in detail here. In other optional embodiments, other types of recurrent neural networks can also be selected.

[0083] S313. Based on the second price sequence, a particle swarm optimization algorithm is used to optimize the core parameters of variational mode decomposition on the basis of the initial food price prediction model, thereby obtaining the food price prediction model and predicting the future food price.

[0084] Specifically, in this embodiment, although the use of variational mode decomposition can decompose the grain price series into multiple first components with different frequency scales, which can improve the prediction accuracy of grain prices, variational mode decomposition has the problem of excessive human intervention in the setting of core parameters, which will affect the accuracy of IMF component acquisition and further affect the accuracy of grain price prediction. If the optimal core parameters of variational mode decomposition can be found, the accuracy of grain price prediction can be further improved, providing more accurate and reliable technical support for subsequent control of grain storage costs and reduction of grain storage risks.

[0085] For more details, see Figure 2 First, a set of core parameters that need to be optimized in variational mode decomposition are used as particles in the particle swarm optimization algorithm. The number of particles is set to 20 and the particle swarm is randomly initialized. At the same time, the maximum number of iterations t of the particle swarm optimization algorithm is set to 30. Then, the second price series is used as the input of the CVMD-RNN model, and multiple second IMF components are obtained through variational mode decomposition. The obtained second IMF components are input into the recurrent neural network to obtain the predicted value of grain prices. Next, the fitness value is calculated according to the predicted value of grain prices and the speed and position of the particles are updated. Finally, it is determined whether the maximum number of iterations has been reached. When the maximum number of iterations has been reached, the combination of the best core parameters is output. Otherwise, the particles are returned to perform variational mode decomposition again until the maximum number of iterations has been reached.

[0086] By using the best core parameters obtained in the CVMD-RNN model, a grain price prediction model can be obtained. In this embodiment, the grain price prediction model is recorded as a VMD-LSTM model. After that, it is only necessary to input the price sequence containing the current grain price into the VMD-LSTM model to realize the prediction of future grain prices.

[0087] Furthermore, the fitness value of the particle swarm optimization algorithm is the absolute value of the difference between the predicted value and the actual value of the grain price. The core parameters that need to be optimized in variational mode decomposition include the number of modes, penalty factor, and bandwidth constraint. The particle swarm optimization algorithm is used to optimize the core parameters of variational mode decomposition, and then the variational mode decomposition is used to obtain the optimal IMF component, which further improves the accuracy of future grain price forecasts.

[0088] S32. Construct a food demand prediction model based on the historical demand sequence, and use the food demand prediction model to predict future food demand.

[0089] Among them, S32 specifically includes the following steps:

[0090] S321. Calculate the average historical demand for grain under the set number of storage cycles.

[0091] Specifically, in this embodiment, the average historical demand satisfies the following relationship:

[0092]

[0093] Where n is the number of storage cycles, is the historical demand data of the i-th storage cycle under the set number of storage cycles, is the average historical demand. The number of storage cycles n can be set according to actual needs, but in order to ensure the accuracy of food demand forecast, , n is generally not less than 5.

[0094] S322: Consider the impact of the period intervals between different grain historical transaction periods on the historical demand, and then build a grain demand forecasting model in combination with the average historical demand. The grain demand forecasting model satisfies the following relationship:

[0095]

[0096] in, For future food needs.

[0097] Specifically, in this embodiment, the food demand forecasting model takes into account the impact of the period interval between food trading cycles on food demand, and can calculate relatively accurate future food demand even when food demand fluctuates drastically, thereby avoiding the problem of food inventory backlog or shortage and reducing food storage risks and costs. In other optional embodiments, the method for predicting future food prices in this embodiment is also applicable to predicting future food demand.

[0098] S4. Use the mathematical fluctuation model to verify the future food price and the future food demand, and store food according to the verification results.

[0099] Wherein, S4 specifically includes the following steps:

[0100] S41. Calculate the quotient of the future food price and the future food demand, and record the calculation result as the volatility index forecast value.

[0101] S42. According to the season corresponding to the fluctuation index prediction value, the fluctuation index prediction value is brought into the corresponding mathematical fluctuation model to calculate the fluctuation decision factor prediction value, and it is determined whether the fluctuation decision factor prediction value is within the fluctuation range.

[0102] Specifically, in this embodiment, the fluctuation range needs to be set according to the type of grain and season, but is generally set between the maximum and minimum values ​​of the mathematical fluctuation model of the season and including the maximum and minimum values, but can also be determined based on expert opinions.

[0103] S43. If the predicted value of the fluctuation determination factor is within the fluctuation range, grain is collected and stored in the warehouse according to the future grain demand.

[0104] Specifically, in this embodiment, if the predicted value of the fluctuation determination factor is within the fluctuation range, it means that the prediction result is not only in line with the historical trend, but also reasonable in actual economic logic. Therefore, grain can be collected and stored in strict accordance with future grain demand, so that the grain storage volume is consistent with the actual market demand, avoiding inventory backlogs or shortages, and reducing grain storage risks and costs.

[0105] S44. If the predicted value of the fluctuation determination factor is not within the fluctuation range, a backup grain storage strategy is adopted to store the grain in a warehouse.

[0106] If the predicted value of the fluctuation determination factor is not within the fluctuation range, it indicates that unknown changes may occur in the future grain market. Therefore, this embodiment uses a backup grain storage strategy to improve the flexibility of grain storage, ensure the stability of grain supply, optimize the use of warehouse resources, and improve grain storage efficiency. The backup grain storage strategy includes the following steps:

[0107] H1. If the future food demand is less than the current food demand, the food will be stored in the warehouse according to the current food demand.

[0108] H2. If the future food demand is greater than the current food demand, the food is stored in the first warehouse according to the current food demand, and the portion of the future food demand that exceeds the current food demand is stored in other warehouses that are closest to the first warehouse.

[0109] S5. Based on the real-time image of the grain warehouse, monitor and warn the grain status based on a deep learning algorithm.

[0110] Among them, S5 specifically includes the following steps:

[0111] S51. Use grain images in different grain states to build a grain status monitoring dataset, and use a convolutional neural network to build an initial monitoring model.

[0112] Specifically, in this embodiment, grain states include normal, damp, moldy, and insect-infested, and a large number of grain pictures representing different grain states are collected using surveillance cameras, and the collected pictures are manually labeled to clarify the grain state corresponding to each picture, and then the pictures are preprocessed as necessary, such as resizing and denoising, and finally the processed pictures are divided into training set, verification set, and test set.

[0113] Furthermore, CNN is used to build an initial monitoring model.

[0114] S52. Use the grain status monitoring data set to complete the training and verification of the initial monitoring model to obtain a grain monitoring model.

[0115] S53. According to the real-time image, use the grain monitoring model to identify the grain status, and issue an alarm when the grain status is abnormal.

[0116] Specifically, in this embodiment, when the grain status is abnormal, a sound alarm is used to prompt relevant personnel to enter the warehouse in time to check the grain and the warehouse environment.

[0117] Furthermore, in other optional embodiments, other alarm methods may be selected, such as flashing lights.

[0118] It should be noted that, in some cases, the actions described in the specification can be performed in a different order and still achieve the desired results. In this embodiment, the order of steps given is only to make the embodiment appear clearer and easier to explain, rather than to limit it.

[0119] In an alternative embodiment, see Figure 3 The present invention also provides a grain storage management system, which uses a grain storage management method provided by the present invention. The system can stably execute the grain storage management method provided by the present invention, improve the efficiency of grain storage, and improve the practicality of the method of the present invention. The system includes a data acquisition module 1, a grain prediction module 2, a monitoring and early warning module 3, a data storage module 4 and a data output module 5.

[0120] The data acquisition module 1 is used to collect historical price data and historical demand data of grain, and simultaneously collect real-time images of grain warehouses.

[0121] Specifically, in this embodiment, the data collection module 1 includes a digital display screen and a surveillance camera. After relevant personnel collect historical price data and historical demand data of grains in different storage cycles on the Internet or in journals, they can input the collected data into the data collection module 1 on the digital display screen. The data collection module 1 transmits the historical price data and historical demand data to the grain forecasting module 2 through the Internet of Things, transmits the real-time image to the monitoring and early warning module 3 through the Internet of Things, and transmits all the collected data to the data storage module 4 through the Internet of Things.

[0122] The food forecasting module 2 is used to construct a historical price sequence and a historical demand sequence, and then construct a mathematical fluctuation model between food prices and food demand; predict future food prices based on the historical price sequence, and predict future food demand based on the historical demand sequence; use the mathematical fluctuation model to verify the future food price and the future food demand, and store food based on the verification results.

[0123] Specifically, in this embodiment, the grain forecasting module 2 receives the data collected by the data collection module 1 through the Internet of Things, and the grain forecasting module 2 is electrically connected to the data storage module 4.

[0124] Furthermore, the food forecasting module 2 specifically executes the contents described in step S2 to step S4.

[0125] The monitoring and early warning module 3 is used to monitor and warn the grain status based on the real-time image of the grain warehouse and based on a deep learning algorithm.

[0126] Specifically, in this embodiment, the monitoring and early warning module 3 is electrically connected to the data storage module 4, and the monitoring and early warning module 3 receives the real-time image collected by the data collection module 1 through the Internet of Things, and then executes the content described in step S5.

[0127] The data storage module 4 is used to store the data obtained by the data acquisition module, the food forecasting module and the monitoring and early warning module.

[0128] Specifically, in this embodiment, the data stored in the data storage module 4 includes historical price data, historical demand data, real-time images, historical price series, historical demand series, fluctuation index series, future food prices, future food demand, food storage plans and food status.

[0129] The data output module 5 is used to output the real-time image, the mathematical fluctuation model, the future food price, the future food demand and the food storage plan.

[0130] Specifically, in this embodiment, the data output module 5 is electrically connected to the data storage module 4, and the data output module 5 includes a digital display screen for displaying real-time images, mathematical fluctuation models, future food prices, future food demand and food storage plans.

[0131] In other optional embodiments, relevant personnel can also set the data output module 5 according to actual needs to query and display other data in the data storage module 4.

[0132] In summary, the present invention firstly constructs a mathematical fluctuation model to reflect the mathematical fluctuation model between grain price and demand, thereby providing a basis for subsequent accurate grain demand forecasting, and then uses the grain price forecasting model and the grain demand forecasting model to respectively forecast future grain prices and future grain demand, and then uses the mathematical fluctuation model to verify the forecasting results, so that warehouse managers can more keenly capture market changes, and then store grain according to different grain storage plans, so that the present method has strong foresight, systematicness and flexibility, avoids the problem of inventory backlog or shortage caused by the disconnection between inventory management strategies and actual market demand, and reduces grain storage risks and costs. In addition, the system can stably execute the grain storage management method provided by the present invention, improve the efficiency of grain storage, and improve the practicality of the method of the present invention.

[0133] Ensure the rationality and accuracy of grain storage, thereby controlling grain storage costs and reducing grain storage risks.

[0134] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the specification of the present invention.

Claims

1. A grain storage management method, characterized in that: The steps include: Collect historical price data and historical demand data of grains in different storage cycles, and collect real-time images of grain warehouses; Constructing a historical price sequence based on the historical price data, and constructing a historical demand sequence based on the historical demand data; The quotient of the grain price and the grain demand in the same storage period is used as a volatility index, and then the historical price series and the historical demand series are used to construct a volatility index series; Dividing the volatility index sequence into a first subsequence, a second subsequence, a third subsequence and a fourth subsequence according to season; Based on the four divided subsequences, the corresponding mathematical fluctuation models are obtained based on Gaussian mixture model and Bayesian information criterion; Dividing the historical price sequence into a first price sequence and a second price sequence, and preprocessing the first price sequence using variational mode decomposition to obtain a first component set; Using the first component in the first component set as the input of the recurrent neural network, and the predicted value of the grain price as the output of the recurrent neural network, and then constructing an initial grain price prediction model based on variational mode decomposition and the recurrent neural network; Based on the second price sequence, a particle swarm optimization algorithm is used to optimize the core parameters of variational mode decomposition on the basis of the initial grain price prediction model, thereby obtaining a grain price prediction model and predicting future grain prices; Calculate the average historical demand for grain for a given number of storage cycles; Considering the impact of the period intervals between different grain historical trading cycles on historical demand, a grain demand forecasting model is constructed in combination with the average historical demand. The grain demand forecasting model satisfies the following relationship: ， Wherein, n is the number of storage cycles, is the historical demand data of the i-th storage cycle under the set number of storage cycles, , To meet future food needs, is the average historical demand; Calculate the quotient of the future food price and the future food demand, and record the calculation result as the volatility index forecast value; According to the season corresponding to the fluctuation index prediction value, the fluctuation index prediction value is brought into the corresponding mathematical fluctuation model to calculate the fluctuation determination factor prediction value, and it is determined whether the fluctuation determination factor prediction value is within the fluctuation range; If the predicted value of the fluctuation determination factor is within the fluctuation range, the grain is collected and stored in the warehouse according to the future grain demand; If the predicted value of the fluctuation determination factor is not within the fluctuation range, the reserve grain storage strategy is adopted to store the grain in the warehouse; According to the real-time images of the grain warehouse, the grain status is monitored and warned based on the deep learning algorithm.

2. A grain storage management method according to claim 1, characterized in that: The standby food storage strategy includes the following steps: If the future food demand is less than the current food demand, the food is stored in the warehouse according to the current food demand; If the future food demand is greater than the current food demand, the food will be stored in the first warehouse according to the current food demand, and the portion of the future food demand that exceeds the current food demand will be stored in other warehouses that are closest to the first warehouse.

3. A grain storage management method according to claim 1, characterized in that: The monitoring and early warning of grain status based on the real-time image of the grain warehouse and the deep learning algorithm comprises the following steps: A grain status monitoring dataset was constructed using grain images in different grain states, and an initial monitoring model was constructed using a convolutional neural network. Using the grain status monitoring data set to complete the training and verification of the initial monitoring model to obtain a grain monitoring model; The grain status is identified according to the real-time image using the grain monitoring model, and an alarm is issued when the grain status is abnormal.

4. A grain storage management system, the grain storage management system using a grain storage management method according to any one of claims 1 to 3, characterized in that: include: A data collection module, which is used to collect historical price data and historical demand data of grain, and collect real-time images of grain warehouses; A grain forecasting module, the grain forecasting module is used to construct a historical price sequence according to the historical price data, and to construct a historical demand sequence according to the historical demand data; The quotient of the grain price and the grain demand in the same storage period is used as a volatility index, and then the historical price series and the historical demand series are used to construct a volatility index series; Dividing the volatility index sequence into a first subsequence, a second subsequence, a third subsequence and a fourth subsequence according to season; Based on the four divided subsequences, the corresponding mathematical fluctuation models are obtained based on Gaussian mixture model and Bayesian information criterion; Dividing the historical price sequence into a first price sequence and a second price sequence, and preprocessing the first price sequence using variational mode decomposition to obtain a first component set; Using the first component in the first component set as the input of the recurrent neural network, and the predicted value of the grain price as the output of the recurrent neural network, and then constructing an initial grain price prediction model based on variational mode decomposition and the recurrent neural network; Based on the second price sequence, a particle swarm optimization algorithm is used to optimize the core parameters of variational mode decomposition on the basis of the initial grain price prediction model, thereby obtaining a grain price prediction model and predicting future grain prices; Calculate the average historical demand for grain for a given number of storage cycles; Considering the impact of the period intervals between different grain historical trading cycles on historical demand, a grain demand forecasting model is constructed in combination with the average historical demand. The grain demand forecasting model satisfies the following relationship: ， Wherein, n is the number of storage cycles, is the historical demand data of the i-th storage cycle under the set number of storage cycles, , To meet future food needs, is the average historical demand; Calculate the quotient of the future food price and the future food demand, and record the calculation result as the volatility index forecast value; According to the season corresponding to the fluctuation index prediction value, the fluctuation index prediction value is brought into the corresponding mathematical fluctuation model to calculate the fluctuation determination factor prediction value, and it is determined whether the fluctuation determination factor prediction value is within the fluctuation range; If the predicted value of the fluctuation determination factor is within the fluctuation range, the grain is collected and stored in the warehouse according to the future grain demand; If the predicted value of the fluctuation determination factor is not within the fluctuation range, the reserve grain storage strategy is adopted to store the grain in the warehouse; A monitoring and early warning module, which is used to monitor and warn the grain status based on the real-time image of the grain warehouse and based on a deep learning algorithm; A data storage module, the data storage module is used to store the data obtained by the data acquisition module, the food forecasting module and the monitoring and early warning module; A data output module, wherein the data output module is used to output the real-time image, the mathematical fluctuation model, the future food price, the future food demand and the food storage plan.

Images