A crop cultivation substrate production management method and system based on a grow-and-feed cycle

Abstract

The application relates to the technical field of data processing, and discloses a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on a crop cultivation substrate production management method and system based on

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a crop cultivation substrate production management method and system based on crop-livestock recycling. BACKGROUND

[0002] With the increasing demand for sustainable agricultural development, the crop-livestock recycling model has attracted much attention due to its efficient use of agricultural waste. Organic vinegar dregs (acetic acid fermentation residues) and silkworm excrement (silkworm feces) are rich in nutrients such as nitrogen, phosphorus, potassium and organic matter, and are excellent raw materials for crop cultivation substrates.

[0003] However, in actual production, the nutrient ratio of raw materials varies greatly depending on the source, and the substrate demand of the planted crops changes dynamically over time. Traditional substrate production management relies on experience to adjust, and lacks dynamic optimization capability, resulting in low resource utilization efficiency and high production cost. SUMMARY

[0004] The present application provides a crop cultivation substrate production management method and system based on crop-livestock recycling, aiming to solve at least one of the above technical problems.

[0005] To achieve the above-mentioned purpose, the present application provides a crop cultivation substrate production management method based on crop-livestock recycling, which comprises the following steps:

[0006] S1: Obtain batch supply information of substrate raw materials, and generate standardized nutrient data of multi-source raw materials;

[0007] S2: Upon receiving a crop planting plan for a target crop cultivation area, extract crop planting planning information from the crop planting plan, analyze the substrate demand and cultivation nutrient requirements of each planting period according to the time sequence, and determine the cultivation predicted demand;

[0008] S3: Based on the standardized nutrient data and the cultivation predicted demand, a multi-objective optimization algorithm is used to solve the initial production strategy of the crop cultivation substrate;

[0009] S4: Taking demand satisfaction and nutrient deviation as state space, and taking production quantity adjustment and ingredient proportion adjustment as action space, a reinforcement learning model based on dynamic deviation penalty is established, and the optimal adjustment strategy is iterated according to the reward value. The adjustment instruction of the optimal adjustment strategy is used to optimize the initial production strategy, and the real-time production strategy of the crop cultivation substrate is generated;

[0010] S5: According to the real-time production strategy, the production management control of the crop cultivation substrate is executed.

[0011] Optionally, the step of obtaining batch supply information of substrate raw materials and generating standardized nutrient data of multi-source raw materials comprises:

[0012] S11: querying batch supply information of different substrate raw material sources; wherein the substrate raw material sources include a plurality of vinegar residue sources and a plurality of silkworm ash sources, and the batch supply information includes a supplier test report of each batch of substrate raw material sources;

[0013] S12: extracting raw material nutrition parameters in the supplier test report of each batch of substrate raw material sources to generate standardized nutrition data containing a plurality of batches of raw material nutrition contents ; wherein i represents the i-th raw material source type, i = 1 for vinegar residue, and i = 2 for silkworm ash; j represents a raw material batch; and k represents the k-th nutrition content.

[0014] Optionally, after receiving a crop planting plan of a target crop cultivation area, crop planting planning information in the crop planting plan is extracted, substrate demand and cultivation nutrition requirements in each planting period are analyzed according to a time sequence, and a cultivation prediction demand step is determined, specifically including:

[0015] S21: when receiving crop planting planning information of a target crop cultivation area, the crop planting planning information is analyzed according to the crop planting planning information; wherein the crop planting planning information contains planting area and planting growth period planning of each crop type;

[0016] S22: according to the planting area and the planting production period of each crop type in different planting periods, the substrate demand and the cultivation nutrition requirements in each planting period are analyzed to determine the cultivation prediction demand.

[0017] Optionally, according to the planting area and the planting production period of each crop type in different planting periods, the substrate demand and the cultivation nutrition requirements in each planting period are analyzed to determine the cultivation prediction demand step, specifically including:

[0018] S221: according to the planting area and the planting production period of each crop type in different planting periods, the substrate demand and the cultivation nutrition requirements in each planting period are analyzed according to the nutrition content demand and the soil basic supply of the crop type in unit planting area and different planting production periods;

[0019] wherein the expression of the substrate demand is specifically:

[0020] ;

[0021] In the formula, represents the substrate demand of each planting period t, represents the planting area of the crop type m in the planting period t, represents the nutrition content demand of the crop type m in unit planting area in the planting period t.

[0022] wherein the expression of the cultivation nutrient requirement is specifically:

[0023] ;

[0024] wherein, represents the cultivation nutrient requirement of each planting period t for the nutrient index k, represents the demand of the crop type m for the nutrient index k per unit planting area at the planting period t, represents the basic supply of the soil for the nutrient index k per unit planting area of the crop type m at the planting period t;

[0025] S222: determining the cultivation predicted demand of the target crop cultivation area according to the substrate demand amount and the cultivation nutrient requirement of the target crop cultivation area at each planting period.

[0026] Optionally, based on the standardized nutrient data and the cultivation predicted demand, an initial production strategy of the crop cultivation substrate is solved by using a multi-objective optimization algorithm, specifically including:

[0027] S31: extracting the nutrient content of several batches of raw materials in the standardized nutrient data and the substrate demand amount in the cultivation predicted demand and the cultivation nutrient requirement ;

[0028] S32: taking the amount of raw materials of the i-th raw material source type of the j-th batch of raw materials used at the planting period t as the decision variable , determining an optimization objective function set of the minimum procurement cost and the maximum nutrient matching degree according to a constraint condition set determined by the total substrate demand amount and the nutrient demand, and solving the initial production strategy of the crop cultivation substrate by using a multi-objective optimization algorithm.

[0029] Optionally, the expression of the constraint condition set determined by the total substrate demand amount and the nutrient demand and the optimization objective function set determined by the minimum procurement cost and the maximum nutrient matching degree is specifically:

[0030] Constraint condition set:

[0031] ;

[0032] ;

[0033] wherein, , represents the nutrient balance interval coefficient;

[0034] Optimization objective function set:

[0035] ;

[0036] ;

[0037] wherein, represents the transportation cost of the unit raw material amount of the i-th raw material source type from the j-th raw material batch, represents the procurement cost of the unit raw material amount of the i-th raw material source type from the j-th raw material batch.

[0038] Optionally, the initial production strategy of the crop cultivation substrate is solved by using a multi-objective optimization algorithm, and specifically includes:

[0039] S321: setting an optimization objective function of minimizing the procurement cost as the main objective function, an optimization objective function of maximizing the nutrient matching degree as the secondary objective function;

[0040] S322: setting a threshold value of the secondary objective function, and converting the secondary objective function into a constraint condition; wherein the expression of the converted constraint condition is specifically:

[0041] ;

[0042] wherein, represents the set nutrient matching degree threshold value;

[0043] S323: aggregating the converted constraint condition with an original constraint condition set determined by the total substrate demand and the nutrient demand, and constructing an optimization model with single objective and multiple constraint conditions according to the aggregated constraint condition set and the optimization objective function of minimizing the procurement cost as the main objective function;

[0044] S324: solving the optimization model by using a heuristic optimization algorithm, and obtaining an optimal plan of the raw material amount of each raw material batch under each raw material source type used in each planting period , and taking the optimal plan as the initial production strategy of the crop cultivation substrate.

[0045] Optionally, a reinforcement learning model based on dynamic deviation penalty is established by taking the demand satisfaction and the nutrient deviation as the state space and taking the production amount adjustment and the ingredient proportion adjustment as the action space, and specifically includes:

[0046] S41: constructing a reinforcement learning environment with the demand satisfaction and the nutrient deviation as the state space, the production amount adjustment and the ingredient proportion adjustment as the action space, and a weighted negative penalty term as the reward function;

[0047] wherein, the expressions of the state space and the action space are specifically:

[0048] ;

[0049] ;

[0050] ;

[0051] ;

[0052] In the formula, Representing the state space, This indicates that the substrate does not meet the requirements. Indicates nutritional imbalance. Represents the action space. This indicates an adjustment in production volume. This indicates an adjustment in the ingredient ratio;

[0053] Specifically, the expression for the reward function based on the weighted negative penalty term is as follows:

[0054] ;

[0055] ;

[0056] In the formula, This represents the reward value during planting period t. , , Indicates the weight of the penalty item. This represents the total procurement and transportation costs during the planting period t;

[0057] S42: Construct a state space As input, with action space probability distribution The output Actor network and the state space and action space As input, with state-action value This is the output Critic network, and the Actor-Critic network is initialized.

[0058] S43: Associate the state space, action space, and reward function in the reinforcement learning environment with the Actor-Critic network, and generate a reinforcement learning model by using the Critic network loss function constructed based on temporal difference error and the Actor network advantage function determined based on the network policy gradient.

[0059] Optionally, the initial production strategy is optimized by using the adjustment instruction of the optimal adjustment strategy to generate a real-time production strategy of the crop cultivation substrate based on the reward value, and the real-time production strategy of the crop cultivation substrate specifically comprises the following steps:

[0060] S44: determining an initial substrate production amount and an initial proportioning ratio according to the amount of the raw material batch of each raw material source type used in each planting period in the initial production strategy;

[0061] wherein the expression of the initial substrate production amount and the initial proportioning ratio is specifically:

[0062] ;

[0063] ;

[0064] wherein, represents the initial substrate production amount, represents the initial proportioning ratio;

[0065] S45: collecting a real-time state of the reinforcement learning environment in each iteration round of each planting period , inputting the real-time state into an Actor network, outputting an action probability distribution through a Softmax function, and sampling to obtain a current adjustment action ;

[0066] S46: correcting the substrate production amount and the proportioning ratio according to the adjustment action, and calculating a corrected raw material amount according to the corrected substrate production amount and the proportioning ratio;

[0067] ;

[0068] ;

[0069] ;

[0070] wherein, represents the corrected raw material amount of each raw material batch of each raw material source type used in each planting period, represents the corrected substrate production amount, represents the corrected proportioning ratio;

[0071] S47: collecting a reward value of the planting period t and a state of the next planting period , updating the Critic network parameters by using a time difference error, updating the Actor network parameters by using a policy gradient, repeating the above iteration process until the policy converges, and outputting an optimal adjustment strategy;​

[0072] S48: output the optimal adjustment strategy as a real-time production strategy of the crop cultivation substrate optimized from the initial production strategy.

[0073] In addition, in order to achieve the above-mentioned purpose, the application also provides a crop cultivation substrate production management system based on the crop breeding cycle, comprising:

[0074] The acquisition module is configured to acquire batch supply information of the substrate raw materials and generate standardized nutrition data of the multi-source raw materials.

[0075] The receiving module is configured to receive a crop planting plan of a target crop cultivation area, extract crop planting planning information in the crop planting plan, analyze substrate demand and cultivation nutrition requirements of each planting period according to a time sequence, and determine cultivation prediction requirements.

[0076] The solving module is configured to solve an initial production strategy of the crop cultivation substrate based on the standardized nutrition data and the cultivation prediction requirements by using a multi-objective optimization algorithm.

[0077] The optimization module is configured to establish a reinforcement learning model based on dynamic deviation punishment by taking demand satisfaction and nutrition deviation as a state space, taking production quantity adjustment and ingredient ratio adjustment as an action space, and iteratively determining an optimal adjustment strategy according to a reward value, and optimize the initial production strategy by using adjustment instructions of the optimal adjustment strategy to generate a real-time production strategy of the crop cultivation substrate.

[0078] The execution module is configured to perform production management control of the crop cultivation substrate according to the real-time production strategy.

[0079] The application has the following beneficial effects: a crop cultivation substrate production management method and system based on the crop breeding cycle are provided, a full-process intelligent management system from raw material proportioning, transportation scheduling to production dynamic regulation and control is constructed through multi-source data fusion, multi-objective optimization and deep reinforcement learning, and optimal decision-making of dynamic and static combination is realized through multi-objective optimization and reinforcement learning based on substrate demand and nutrition demand in demand prediction. BRIEF DESCRIPTION OF DRAWINGS

[0080] Figure 1A flowchart of a crop cultivation substrate production management method based on the seed-crop cycle according to an embodiment of the present application is shown in FIG. 1.

[0081] Figure 2 A structural diagram of a crop cultivation substrate production management system based on the seed-crop cycle according to an embodiment of the present application is shown in FIG. 2. DETAILED DESCRIPTION

[0082] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application.

[0083] An embodiment of the present application provides a crop cultivation substrate production management method based on the seed-crop cycle, which is described below with reference to Figure 1 , Figure 1 A flowchart of a crop cultivation substrate production management method based on the seed-crop cycle according to an embodiment of the present application is shown in FIG. 1.

[0084] In this embodiment, a crop cultivation substrate production management method based on the seed-crop cycle includes the following steps:

[0085] S1: Obtain batch supply information of substrate raw materials, and generate standardized nutrition data of multiple sources of raw materials;

[0086] S2: After receiving a crop planting plan of a target crop cultivation area, extract crop planting planning information in the crop planting plan, analyze the substrate demand and cultivation nutrition requirements of each planting period according to the time sequence, and determine cultivation prediction requirements;

[0087] S3: Based on the standardized nutrition data and the cultivation prediction requirements, a multi-objective optimization algorithm is used to solve the initial production strategy of the crop cultivation substrate;

[0088] S4: A reinforcement learning model based on dynamic deviation punishment is established by taking demand satisfaction and nutrition deviation as a state space and taking production quantity adjustment and ingredient proportion adjustment as an action space, and an optimal adjustment strategy is iteratively adjusted according to a reward value, and the initial production strategy is optimized by using the adjustment instruction of the optimal adjustment strategy to generate a real-time production strategy of the crop cultivation substrate;

[0089] S5: According to the real-time production strategy, the production management control of the crop cultivation substrate is performed.

[0090] It should be noted that organic vinegar dregs (acetic acid fermentation residues) and silkworm excrement (silkworm feces) are rich in nutrients such as nitrogen, phosphorus, potassium and organic matter, and are high-quality raw materials for crop cultivation substrates. However, in actual production, the nutrient ratio of raw materials varies greatly depending on the source, and the substrate demand of the planted crops changes dynamically over time. Traditional substrate production management relies on experience to adjust, lacks dynamic optimization capability, resulting in low resource utilization efficiency and high production cost.

[0091] To solve the above problems, the embodiment aims at the precise production demand of resource utilization of agricultural waste (vinegar dregs, silkworm excrement) in the scenario of crop cultivation and breeding, and establishes a crop cultivation substrate production management method covering the whole link through a five-stage closed-loop process of raw material data standardization, cultivation demand prediction, multi-objective optimization initial strategy, reinforcement learning dynamic adjustment and production control landing. The production strategy is upgraded from static planning to dynamic adaptation, providing a standardized technical framework for efficient resource utilization of agricultural waste.

[0092] In the preferred embodiment, the batch supply information of the substrate raw material is obtained, and the standardized nutrient data of the multi-source raw material is generated, specifically including:

[0093] S11: query batch supply information of different substrate raw material sources; wherein the substrate raw material sources include a plurality of vinegar dregs sources and a plurality of silkworm excrement sources, and the batch supply information includes a supplier test report of each batch of substrate raw material source;

[0094] S12: extract the raw material nutrient parameters in the supplier test report of each batch of substrate raw material source, and generate standardized nutrient data containing a plurality of batches of raw material nutrient contents ; wherein i represents the i-th raw material source type, i=1 for vinegar dregs, i=2 for silkworm excrement; j represents the raw material batch; k represents the k-th nutrient content.

[0095] In the embodiment, in view of the heterogeneity problem of nutrient data caused by source difference (different vinegar factories, silkworm bases) and batch fluctuation of raw materials such as vinegar dregs and silkworm excrement, the standardized nutrient parameters are extracted by directional collection of multi-source batch supply information, and a unified format of raw material nutrient database is generated, providing a reliable data basis for subsequent batching optimization. Therefore, the present application converts the dispersed report data into structured and calculable standardized nutrient data set by obtaining the supplier test report of each batch from different substrate raw material sources, ensures the comparability of raw material data from different sources and batches, makes the nutrient characteristics of multi-source raw materials quantifiable and comparable, provides a unified data benchmark for subsequent precise batching according to nutrient demand, and avoids batching deviation caused by data heterogeneity.

[0096] In the preferred embodiment, after receiving the crop planting plan of the target crop cultivation area, the crop planting planning information in the crop planting plan is extracted, the substrate demand and cultivation nutrient requirement of each planting period are analyzed according to the time sequence, and the cultivation prediction demand step is determined, specifically comprising:

[0097] S21: when receiving the crop planting planning information of the target crop cultivation area, according to the crop planting planning information; wherein the crop planting planning information contains the planting area and planting growth period planning of each crop type;

[0098] S22: according to the planting area of each crop type in different planting periods and the planting production period, the substrate demand and cultivation nutrient requirement of each planting period are analyzed, and the cultivation prediction demand is determined.

[0099] In this embodiment, first, the crop planting plan of the target area is received (which can be an electronic document or system input data), and two types of information are extracted: a crop type list (such as tomatoes, cucumbers, and lettuce) and a single crop planting plan (including the total planting area of each crop, the allocation of the planting area in different periods, the division of the growth period (such as the seedling period, the nutrient growth period, and the reproductive growth period), the expected harvest period, etc.), ensuring that the information covers the three dimensions of space, time, and biological characteristics. After that, based on the growth law of crops, the substrate demand (amount) and nutrient supply demand (NPK, etc.) in different growth periods are significantly different (such as the seedling period requiring less substrate amount but sufficient nitrogen, and the result period requiring a large amount and rich in phosphorus and potassium). Combined with the actual planting area of each crop in each planting period and the growth period, through the time period, crop, and demand correlation model fitted by historical production data, the total substrate demand and cultivation demand of each nutrient index in each period are calculated, and finally the cultivation prediction demand list in each period is summarized.

[0100] Therefore, according to the planting planning file of the target crop cultivation area, the core information of crop types, planting area, and growth period division is extracted, and the substrate consumption characteristics and nutrient demand law of different growth stages of crops are combined. Through the analysis and quantification of substrate demand and nutrient supplement requirements in different periods, the abstract planting plan is converted into specific production targets, thereby converting the vague planting task into precise production indexes, clearly defining the amount of substrate production in each period and the nutrient standard, providing clear targets for subsequent production strategy formulation, and avoiding the accumulation of raw materials or insufficient supply due to unclear demands.

[0101] On this basis, according to the planting area of each crop type in different planting periods and the planting production period, the substrate demand and cultivation nutrient requirement of each planting period are analyzed, and the cultivation prediction demand step is determined, specifically comprising:

[0102] S221: According to the planting area and the planting production period of each crop type in different planting periods, the substrate demand and the cultivation nutrient requirement of each planting period are analyzed based on the nutrient content demand and the soil basic supply of the crop type in unit planting area and different planting production periods.

[0103] The expression of the substrate demand is specifically:

[0104]

[0105] In the formula, represents the substrate demand of each planting period t, represents the planting area of crop type m in planting period t, represents the nutrient content demand of unit planting area of crop type m in planting period t.

[0106] The expression of the cultivation nutrient requirement is specifically:

[0107]

[0108] In the formula, represents the cultivation nutrient requirement of each planting period t for nutrient index k, represents the demand of unit planting area of crop type m in planting period t for nutrient index k, represents the basic supply of soil for unit planting area of crop type m in planting period t for nutrient index k.

[0109] S222: According to the substrate demand and the cultivation nutrient requirement of the target crop cultivation area in each planting period, the cultivation prediction demand of the target crop cultivation area is determined.

[0110] In this embodiment, based on the crop planting area, the unit demand in the growth period and the soil basic supply capacity, the substrate demand and the nutrient supplement requirement of each planting period are accurately calculated through a quantitative formula, and the quantification and calculable expression of the cultivation prediction demand are realized. It should be noted that due to the consideration of the soil basic supply capacity, when determining the cultivation nutrient demand, the demand and supply gap of single crop and single nutrient are calculated first (if the supply is greater than or equal to the demand, the gap is 0, and no substrate supplement is needed), and then all crop gaps are added to obtain the supplement requirement of nutrient k in period t, to ensure the accuracy and operability of the demand index, so that the subsequent production strategy optimization has clear data basis.

[0111] In a preferred embodiment, based on the standardized nutrient data and the cultivation prediction demand, a multi-objective optimization algorithm is used to solve the initial production strategy of crop cultivation substrate, specifically including:

[0112] ​​S31: Extracting the nutrient content of several batches of raw materials in the standardized nutrition data and the substrate demand amount in the cultivation prediction demand and the cultivation nutrient requirement ;

[0113] S32: Using the amount of raw material of the i-th raw material source type of the j-th raw material batch at the planting period t as the decision variable , the constraint condition set determined by the total substrate demand amount and the nutrient requirement, and the optimization objective function set determined by the minimum purchase cost and the maximum nutrient matching degree, a multi-objective optimization algorithm is used to solve the initial production strategy of the crop cultivation substrate.

[0114] Further, the expressions of the constraint condition set determined by the total substrate demand amount and the nutrient requirement, and the optimization objective function set determined by the minimum purchase cost and the maximum nutrient matching degree are as follows:

[0115] Constraint condition set:

[0116] ;

[0117] ;

[0118] In the formula, , denotes the nutrient balance interval coefficient;

[0119] Optimization objective function set:

[0120] ;

[0121] ;

[0122] In the formula, denotes the transportation cost of the unit amount of raw material of the i-th raw material source type purchased from the j-th raw material batch, denotes the purchase cost of the unit amount of raw material of the i-th raw material source type purchased from the j-th raw material batch.

[0123] In this embodiment, first, the key parameters are extracted from the standardized nutrition data set, the substrate demand amount at period t and the nutrient k supplement requirement at period t are extracted from the cultivation prediction demand, and the raw material purchase and transportation cost data (from supplier quotes) are supplemented at the same time, forming a complete input data set of the optimization model, and through outlier detection (for example, if it is outside the normal range of the same type of raw material, it is excluded), the data reliability is ensured. After that, set That is, the amount of raw material of the i-th type (vinegar lees / cast) of the j-th batch used in the planting period t is a decision variable, based on the actual production demand, the total amount of substrate constraint (to ensure that the amount is up to standard), nutrient balance constraint (to ensure that the nutrients are adapted), and other core constraints are constructed, through setting double target: the total cost of procurement and transportation minimization (to control the economy), the maximum degree of nutrient matching (to control the quality of the product), a multi-objective optimization algorithm is used to solve the model, and the globally optimal initial production plan is generated, under the premise of meeting the substrate amount and nutrient demand, the precise control of procurement and transportation cost is realized.

[0124] Therefore, based on the standardized raw material nutrient data and the quantitative cultivation prediction demand, the amount of raw material used in each period and each batch is used as the core decision variable, a mathematical model integrating constraint conditions and optimization objectives is constructed, and a multi-objective optimization algorithm is used to solve it, and an initial production strategy considering cost and nutrient adaptation is obtained.

[0125] On this basis, the steps of solving the initial production strategy of crop cultivation substrate by using a multi-objective optimization algorithm specifically include:

[0126] S321: the optimization objective function of minimizing the procurement cost As the main objective function, the optimization objective function of maximizing the nutrient matching degree As the secondary objective function;

[0127] S322: set the threshold of the secondary objective function, and convert the secondary objective function into a constraint condition; wherein the expression of the converted constraint condition is specifically:

[0128] ;

[0129] In the formula, The set nutrient matching degree threshold is represented by

[0130] S323: the converted constraint condition is summarized with the original constraint condition set determined by the total demand of the substrate and the nutrient demand, and the optimization model of single objective and multiple constraints is constructed according to the summarized constraint condition set and the optimization objective function of minimizing the procurement cost as the main objective function .

[0131] S324: a heuristic optimization algorithm is used to solve the optimization model, and the optimal planning of the amount of raw material of each batch of raw material using each raw material source type in each planting period is obtained, and the optimal planning is used as the initial production strategy of crop cultivation substrate.

[0132] In this embodiment, firstly, the minimum purchase cost is set as the main target and the maximum nutrition matching degree is set as the secondary target according to the actual demand of agricultural production, the optimization priority is determined, the minimum acceptable threshold of the secondary target is set (determined based on crop growth experiments, usually 0.85-0.9, that is, the nutrition matching degree is greater than or equal to 85%-90%) to avoid the dispersion of the solution set of multi-objective optimization, the original objective function is converted into a constraint condition to ensure that the nutrition adaptability of the solution result is not lower than the bottom line, and the converted nutrition matching degree constraint is combined with the original total substrate constraint and nutrition balance constraint to form a complete constraint condition set; the main target is taken as the optimization target to construct a single-objective multi-constraint mathematical model for solving.

[0133] Therefore, by means of the method of primary and secondary target division, secondary target conversion into a constraint, and single-target model solving, the complexity of multi-objective optimization is simplified, the heuristic algorithm (such as genetic algorithm) is used for efficient solving, the initial production strategy meeting the production priority is obtained, the engineering efficient solution of the multi-objective optimization problem is realized, and the production cost is maximally reduced under the premise of ensuring the bottom line of the nutrition matching degree.

[0134] In the preferred embodiment, a reinforcement learning model based on dynamic deviation penalty is established by taking demand satisfaction and nutrition deviation as the state space and taking production quantity adjustment and ingredient proportion adjustment as the action space, and the specific steps include:

[0135] S41: constructing a reinforcement learning environment with demand satisfaction and nutrition deviation as the state space, production quantity adjustment and ingredient proportion adjustment as the action space, and a weighted negative penalty term as the reward function;

[0136] Wherein, the expressions of the state space and the action space are as follows:

[0137] ;

[0138] ;

[0139] ;

[0140] ;

[0141] In the formula, represents the state space, represents the substrate unsatisfied demand, represents the nutrition deviation, represents the action space, represents the production quantity adjustment, represents the ingredient proportion adjustment;

[0142] Wherein, the expression of the reward function based on the weighted negative penalty term is as follows:

[0143] ;

[0144] ;

[0145] In the formula, This represents the reward value during planting period t. , , Indicates the weight of the penalty term. This represents the total procurement and transportation costs during the planting period t;

[0146] S42: Construct a state space As input, with action space probability distribution The output Actor network and the state space and action space As input, with state-action value This is the output Critic network, and the Actor-Critic network is initialized.

[0147] S43: Associate the state space, action space, and reward function in the reinforcement learning environment with the Actor-Critic network, and generate a reinforcement learning model by using the Critic network loss function constructed based on temporal difference error and the Actor network advantage function determined based on the network policy gradient.

[0148] It should be noted that this invention employs an Actor-Critic dual-network architecture, responsible for action generation and value evaluation respectively. The Actor network consists of: a state space... As input, with action space probability distribution The output is a network structure containing an input layer, hidden layers, and a Softmax output layer; the Critic network is a network structure with a state space... and action space As input, with state-action value The output is a network structure containing an input layer, hidden layers, and a Softmax output layer.

[0149] In practical applications, the temporal difference error of the Critic network is used to measure the value prediction bias of the Critic network, and its expression is: , This is the discount factor. The Critic network constructs its loss function during updates. The loss is minimized using gradient descent. The Critic network updates based on the policy gradient method, aiming to maximize the expected reward. The expression for the policy gradient is: wherein, is the advantage function (f is the state value, computed by the Critic network), the parameters are updated in the direction of the gradient.

[0150] In this embodiment, for real-time fluctuations (such as demand changes, raw material nutrient deviations) that the initial production strategy cannot cope with, a reinforcement learning model based on the Actor-Critic architecture is constructed, taking state feedback (demand deviation, nutrient deviation) in the production process as input, and adjusting the action (production quantity, ingredient ratio) and reward feedback (cost, deviation penalty) to realize dynamic optimization of the strategy.

[0151] On this basis, the optimal adjustment strategy is iteratively adjusted based on the reward value, and the adjustment instructions of the optimal adjustment strategy are used to optimize the initial production strategy to generate a real-time production strategy for crop cultivation substrate, specifically including:

[0152] S44: According to the initial production strategy, the initial substrate production quantity and the initial ingredient ratio are determined by using the initial production quantity of each raw material batch of each raw material source type in each planting period.

[0153] wherein, the expression of the initial substrate production quantity and the initial ingredient ratio is:

[0154] ;

[0155] ;

[0156] wherein, represents the initial substrate production quantity, represents the initial ingredient ratio;

[0157] S45: In each iteration round of each planting period, the real-time state of the reinforcement learning environment is collected , and the real-time state is input into the Actor network to output the action probability distribution through the Softmax function, and the current adjustment action is obtained by sampling.

[0158] S46: According to the adjustment action, the substrate production quantity and the ingredient ratio are corrected, and the corrected raw material quantity is calculated according to the corrected substrate production quantity and the ingredient ratio.

[0159] ;

[0160] ;

[0161] ; ​

[0162] wherein, represents the corrected raw material usage of each raw material batch under each raw material source type for each planting period, represents the corrected substrate production amount, represents the corrected batching ratio;

[0163] S47: After performing the correction action, collect the reward value of the planting period t and the state of the next planting period , update the Critic network parameters using the time difference error, update the Actor network parameters using the policy gradient, repeat the above iteration process until the policy converges, and output the optimal adjustment policy;

[0164] S48: Output the optimal adjustment policy as the real-time production strategy of the optimized agricultural substrate based on the initial production strategy.

[0165] In this embodiment, first, the initial raw material usage of each period is extracted from the initial production strategy, which is converted into two core reference parameters: the initial substrate production amount and the initial batching ratio; then by setting the number of iterations, at each iteration round, real-time data is obtained through the production monitoring system, and the state vector is calculated The state vector is input into the trained Actor network, the network output is converted into action probability distribution through the Softmax function, and the current adjustment action is sampled; thereafter, according to the principle of total amount x ratio, taking the initial parameters as the reference, combined with the adjustment work, the corrected parameters are calculated to ensure that the corrected usage meets the total amount demand and adapts to the new batching ratio; after performing the correction action, the reward value of the current period and the state of the next period are collected, the Critic network parameters are updated using the time difference error , and the Actor network parameters are updated using the policy gradient ; when the reward value of 100 consecutive rounds fluctuates by ≤3%, it is determined that the policy converges and the optimal adjustment policy is output ; finally, based on the optimal adjustment policy, specific adjustment instructions (production amount adjustment , batching ratio adjustment ) are generated for each planting period t, and the usage is corrected to obtain the real-time raw material usage.

[0166] Therefore, based on the initial production strategy, the closed-loop optimization of state perception, action decision, reward feedback and parameter update is realized through multiple iterations of the reinforcement learning model, and the real-time production strategy is output after the strategy converges, and the dynamic correction of the initial plan is completed. By multi-objective optimization planning the initial production strategy, scientific configuration and intelligent management of multi-source substrate production raw materials are realized, and the initial production strategy is dynamically optimized by using reinforcement learning, which can adapt to the influence of different actual factors on the production strategy, solve the problem of handling the heterogeneity and supply uncertainty of raw materials, realize the optimization of prediction accuracy and resource utilization efficiency, balance the cost and improve the substrate adaptability, and promote the development of intelligent agricultural planting and breeding circulation.

[0167] Reference Figure 2 , Figure 2 FIG. 1 is a structural schematic diagram of a crop cultivation substrate production management system based on planting and breeding circulation according to an embodiment of the present application.

[0168] As shown in FIG. 2, the crop cultivation substrate production management system based on planting and breeding circulation according to an embodiment of the present application comprises: Figure 2

[0169] The acquisition module 10 is configured to acquire batch supply information of the substrate raw materials, and generate standardized nutrition data of the multi-source raw materials.

[0170] The receiving module 20 is configured to receive a crop planting plan of a target crop cultivation area, extract crop planting planning information in the crop planting plan, analyze the substrate demand and cultivation nutrition requirements of each planting period according to the time sequence, and determine the cultivation prediction demand.

[0171] The solving module 30 is configured to solve an initial production strategy of the crop cultivation substrate based on the standardized nutrition data and the cultivation prediction demand, and adopt a multi-objective optimization algorithm.

[0172] The optimization module 40 is configured to take the demand satisfaction and the nutrition deviation as a state space, take the production quantity adjustment and the ingredient proportion adjustment as an action space, establish a reinforcement learning model based on dynamic deviation punishment, iteratively optimize the adjustment strategy according to the reward value, and optimize the initial production strategy by using the adjustment instruction of the optimal adjustment strategy to generate a real-time production strategy of the crop cultivation substrate.

[0173] The execution module 50 is configured to execute production management control of the crop cultivation substrate according to the real-time production strategy.

[0174] Other embodiments or specific implementation manners of the crop cultivation substrate production management system based on planting and breeding circulation according to the present application can refer to the above-mentioned method embodiments, which will not be described here again.

[0175] ​It is to be understood that the terms "one embodiment", "another embodiment", "other embodiments", "first embodiment", "second embodiment", etc. as may be found in the specification and / or in the claims, indicate that the alternative is included in at least one embodiment. These terms only specify the scope of claimable subject matter; and do not necessarily affect the scope of the application. Further, these terms only indicate particular embodiments of the applications. Other embodiments of the present application can be derived from the description, experimental results and / or the claims, without departing from the scope of the present application.

[0176] It is to be understood that the terms "including", "comprising", "consisting" and "consisting essentially of" or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0177] The above only describes the preferred embodiments of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation, or direct or indirect application in other related technical fields, which is made according to the content of the specification and drawings of the present application, is also included in the patent protection scope of the present application.

Claims

1. A method for production management of crop cultivation substrate based on crop-livestock cycle, characterized in that, The method includes the following steps: S1: Obtain batch supply information of matrix raw materials and generate standardized nutritional data of multi-source raw materials; S2: Upon receiving the crop planting plan for the target crop cultivation area, extract the crop planting planning information from the plan, analyze the substrate requirements and nutrient requirements for each planting period based on time series analysis, and determine the predicted cultivation needs; specifically including: S21: Upon receiving the crop planting plan for the target crop cultivation area, extract the crop planting planning information from the crop planting plan; wherein, the crop planting planning information includes the planting area and planting growth period plan for each crop type; S22: Based on the planting area and growth stage of each crop type at different planting periods, analyze the substrate requirements and nutrient requirements for each planting period to determine the predicted cultivation needs; specifically including: S221: Based on the planting area and growth stage of each crop type at different planting periods, and according to the nutrient content requirements and soil supply of each crop type per unit planting area and at different planting growth stages, analyze the substrate requirements and cultivation nutrient requirements for each planting period. S222: Determine the predicted cultivation requirements of the target crop cultivation area based on the substrate requirements and cultivation nutrient requirements of the target crop cultivation area at each planting period; S3: Based on the standardized nutrient data and the predicted cultivation requirements, a multi-objective optimization algorithm is used to solve the initial production strategy for crop cultivation substrate; specifically including: S31: Extract the nutrient content of several batches of raw materials from the standardized nutrient data. and the substrate requirement in the predicted cultivation needs Nutritional requirements for cultivation Where i represents the type of the i-th raw material source, i=1 is vinegar residue, i=2 is silkworm excrement; j represents the batch of raw material; k represents the nutrient content of the k-th type. This represents the substrate requirement for each planting period t; This represents the cultivation nutrient requirements for each planting period t relative to nutrient index k. S32: The decision variable is the amount of raw material from the j-th batch of the i-th raw material source type used during planting period t. Based on the set of constraints determined by the total substrate demand and nutrient demand, and the set of optimization objective functions determined by minimizing procurement costs and maximizing nutrient matching, a multi-objective optimization algorithm is used to solve the initial production strategy of crop cultivation substrate. S4: Using demand satisfaction and nutritional deviation as the state space, and production volume adjustment and ingredient ratio adjustment as the action space, a reinforcement learning model based on dynamic deviation penalty is established. The optimal adjustment strategy is iterated based on the reward value. The adjustment instructions of the optimal adjustment strategy are used to optimize the initial production strategy and generate a real-time production strategy for crop cultivation substrate. S5: Implement production management and control of crop cultivation substrate according to the real-time production strategy.

2. The method for production management of crop cultivation substrate based on crop-livestock cycle as described in claim 1, characterized in that, The steps for obtaining batch supply information of matrix raw materials and generating standardized nutritional data for multi-source raw materials include: S11: Query batch supply information for different matrix raw material sources; wherein, the matrix raw material sources include several vinegar residue sources and several silkworm excrement sources, and the batch supply information includes the supplier's test report for each batch of matrix raw material source; S12: Extract the nutritional parameters of the raw materials from the supplier's test reports for each batch of matrix raw materials, and generate a list containing the nutritional content of several batches of raw materials. Standardized nutritional data; where i represents the type of the i-th raw material source, i=1 is vinegar residue, i=2 is silkworm excrement; j represents the batch of raw material; k represents the content of the k-th nutrient.

3. The method for production management of crop cultivation substrate based on crop-livestock cycle as described in claim 2, characterized in that, In step S221: The expression for the required matrix is ​​as follows: ； In the formula, This represents the substrate requirement for each planting period t. This represents the planting area of ​​crop type m during planting period t. This represents the nutrient content requirement per unit planting area for crop type m during planting period t. The expression for the cultivation nutrient requirements is as follows: ； In the formula, This represents the cultivation nutrient requirements for nutrient index k at each planting period t. This represents the nutrient requirement k per unit planting area for crop type m during planting period t. This represents the basic supply of soil to nutrient index k per unit planting area for crop type m during planting period t.

4. The method for production management of crop cultivation substrate based on crop-livestock cycle as described in claim 3, characterized in that, The expressions for the set of constraints determining the total substrate demand and nutrient demand, and the set of optimization objective functions determined by minimizing procurement costs and maximizing nutrient matching, are as follows: Constraint set: ； ； In the formula, , Indicates the coefficient for the nutritional balance range; Optimize the objective function set: ； ； In the formula, This represents the transportation cost per unit quantity of raw material from the j-th batch, for the i-th source type. This represents the purchase cost per unit quantity of raw material from the j-th batch, for the ith raw material source type.

5. The method for production management of crop cultivation substrate based on crop-livestock cycle as described in claim 4, characterized in that, A multi-objective optimization algorithm is used to solve the initial production strategy steps for crop cultivation substrates, specifically including: S321: The optimization objective function that minimizes procurement costs. As the primary objective function, the optimization objective function is to maximize the nutrient matching degree. As a secondary objective function; S322: Set a threshold for the secondary objective function, and convert the secondary objective function into a constraint condition; wherein the expression of the converted constraint condition is as follows: ； In the formula, This indicates the set nutritional matching threshold. S323: Summarize the transformed constraints with the original constraint set determined by the total matrix requirement and nutrient requirement, and then optimize based on the summarized constraint set and the optimization objective function of minimizing procurement cost as the main objective function. Construct a single-objective optimization model with multiple constraints; S324: Solve the optimization model using a heuristic optimization algorithm to obtain the raw material usage for each batch of each raw material source type at each planting period. The optimal plan is used as the initial production strategy for crop cultivation substrate.

6. The method for production management of crop cultivation substrate based on crop-livestock cycle as described in claim 5, characterized in that, Using demand satisfaction and nutritional deviation as the state space, and production quantity adjustment and ingredient ratio adjustment as the action space, the steps for establishing a reinforcement learning model based on dynamic deviation penalty include: S41: Construct a reinforcement learning environment with demand satisfaction and nutritional deviation as the state space, production quantity adjustment and ingredient ratio adjustment as the action space, and a reward function based on a weighted negative penalty term; The expressions for the state space and action space are as follows: ； ； ； ； In the formula, Representing the state space, This indicates that the substrate does not meet the requirements. Indicates nutritional imbalance. Represents the action space, This indicates an adjustment in production volume. This indicates an adjustment in the ingredient ratio; Specifically, the expression based on the weighted negative penalty term as the reward function is as follows: ； ； In the formula, This represents the reward value during planting period t. , , Indicates the weight of the penalty term. This represents the total procurement and transportation costs during the planting period t; S42: Construct a state space As input, with action space probability distribution The output Actor network and the state space and action space As input, with state-action value This is the output Critic network, and the Actor-Critic network is initialized. S43: Associate the state space, action space, and reward function in the reinforcement learning environment with the Actor-Critic network, and generate a reinforcement learning model by using the Critic network loss function constructed based on temporal difference error and the Actor network advantage function determined based on the network policy gradient.

7. The method for production management of crop cultivation substrate based on crop-livestock cycle as described in claim 6, characterized in that, The steps of iteratively adjusting the optimal strategy based on reward values, optimizing the initial production strategy using the adjustment instructions of the optimal adjustment strategy, and generating a real-time production strategy for crop cultivation substrate specifically include: S44: Based on the initial production strategy, use the raw material usage for each batch of each raw material source type for each planting period. Determine the initial substrate production quantity and initial ingredient ratio; The expressions for the initial matrix production amount and the initial ingredient ratio are as follows: ； ； In the formula, This indicates the initial substrate production quantity. Indicates the initial ingredient ratio; S45: During each iteration of the planting period, collect the real-time state of the reinforcement learning environment. Real-time status Inputting the Actor network, the Softmax function outputs an action probability distribution, and the current adjustment action is obtained by sampling. ; S46: Adjust the matrix production volume and ingredient ratio according to the adjustment action, and calculate the adjusted raw material consumption based on the adjusted matrix production volume and ingredient ratio; ； ； ； In the formula, This indicates the adjusted raw material usage for each batch under each raw material source type at each planting stage. This indicates the corrected matrix production volume. This indicates the revised ingredient ratio; S47: After executing the correction action, collect the reward value for the planting period t. and the status of the next planting period The parameters of the Critic network are updated using the temporal difference error, and the parameters of the Actor network are updated using the policy gradient. The above iterative process is repeated until the policy converges, and the optimal adjustment policy is output. S48: Output the optimal adjustment strategy as a real-time production strategy for the crop cultivation substrate after optimizing the initial production strategy.

8. A crop cultivation substrate production management system based on crop-livestock cycle, characterized in that, The system is used to execute the crop cultivation substrate production management method based on crop-livestock cycle as described in any one of claims 1-7, and the system includes: The acquisition module is used to acquire batch supply information of matrix raw materials and generate standardized nutritional data of multi-source raw materials; The receiving module is used to extract the crop planting planning information from the crop planting plan when it receives the crop planting plan for the target crop cultivation area, and to determine the cultivation forecast requirements based on the time series analysis of the substrate requirements and cultivation nutrient requirements for each planting period. The solution module is used to solve the initial production strategy of crop cultivation substrate based on the standardized nutrient data and the predicted cultivation requirements, using a multi-objective optimization algorithm. The optimization module is used to establish a reinforcement learning model based on dynamic deviation penalty, with demand satisfaction and nutritional deviation as the state space and production volume adjustment and ingredient ratio adjustment as the action space. It iterates the optimal adjustment strategy based on the reward value, optimizes the initial production strategy using the adjustment instructions of the optimal adjustment strategy, and generates a real-time production strategy for crop cultivation substrate. The execution module is used to perform production management and control of crop cultivation substrate according to the real-time production strategy.

Images