Multi-storey building pig raising feed conveying scheduling method and system based on multi-objective optimization

Abstract

The invention relates to the technical field of intelligent breeding and logistics optimization control, and solves the technical problems of high energy consumption, unstable efficiency, unbalanced distribution, lack of an intelligent scheduling mechanism and the like in feed conveying of a multi-storey pig farm. The method comprises the following steps: acquiring static parameters (physical characteristics of feed, physical attributes of a conveying system and a pig house structure) and dynamic parameters (real-time feeding requirements, equipment and material states and external environment factors); establishing a multi-objective optimization model of a collaborative optimization energy consumption model E (x), a time model T (x) and a conveying balance degree model U (x); solving by adopting a genetic algorithm with a special design crossover and mutation operator to obtain an optimal scheduling scheme; in the execution process, a closed-loop self-learning calibration mechanism is started, actual power is measured through a current sensor, and when the deviation between predicted energy consumption and actual energy consumption exceeds a preset threshold value, efficiency parameters in the energy consumption model are reversely corrected through a gradient descent method; the weight coefficients of the three models are dynamically adjusted according to the real-time electricity price and the inventory state. The system adopts a three-layer architecture of a perception and data acquisition layer, a decision and control core layer and an execution layer. According to the method, multi-target collaborative optimization and intelligent adaptive scheduling are realized, the total energy consumption is effectively reduced, the transmission time is shortened, the distribution balance degree is improved, and the energy consumption prediction accuracy is remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent breeding and logistics optimization control technology, and specifically relates to a method and system for synergistic optimization of feed delivery energy efficiency and balance in multi-story pig farms, which is applied to intelligent precision feeding systems. Background Technology

[0002] With the advancement of the modernization of my country's animal husbandry, the multi-story pig farming model has gradually become popular due to its advantages such as high land utilization and strong environmental controllability.

[0003] However, multi-story pig farms still face many obvious technical challenges in feed delivery:

[0004] First, the problem of high energy consumption is prominent.

[0005] Traditional multi-layer vertical conveying systems need to overcome gravity to vertically lift large amounts of feed, and frequent equipment start-up and shutdown operations result in serious waste of power resources.

[0006] Secondly, the conveying efficiency is unstable.

[0007] During peak feeding periods, conflicts can easily occur between multiple transport routes, leading to significant delays in feeding times in some areas and resulting in overall inefficiency.

[0008] Secondly, the uneven distribution is a serious problem.

[0009] Due to differences in transport paths and equipment performance, there are significant deviations in feed flow between floors, which seriously affect breeding results and feed utilization.

[0010] Finally, there is a lack of intelligent scheduling mechanisms.

[0011] Existing systems mostly employ simple strategies based on fixed sequences and static parameters, which cannot adapt to dynamic conditions such as changes in pig age, feed batch replacements, and real-time electricity price fluctuations.

[0012] The prevalence of these problems stems from the following deep-seated technical flaws in the design philosophy and technical architecture of existing feed delivery systems:

[0013] First, the optimization objective is singular and lacks real-time scheduling capabilities.

[0014] Existing systems typically only consider a single objective (such as time), failing to achieve dynamic and coordinated optimization of energy consumption, time, and balance, and lacking the ability to respond quickly to real-time changes in feeding requirements and equipment status.

[0015] Second, the problem of staticizing energy consumption models.

[0016] Existing technologies generally establish static energy consumption models based on equipment factory parameters, without considering actual operating conditions such as wear and tear during long-term operation and changes in the characteristics of different batches of feed. This results in a huge discrepancy between model predictions and actual operation, making it impossible to achieve true energy savings.

[0017] Third, the special technological constraints of raising pigs in multi-story buildings were not fully considered.

[0018] Existing systems are insufficient in handling complex constraints in multi-story pig farming, such as the dominant energy consumption of vertical lifting, strict feeding time window restrictions, and requirements for preventing cross-contamination between different feeds, making it difficult to formulate a truly feasible optimal solution.

[0019] Therefore, existing technologies have not yet provided an intelligent scheduling method that can combine multi-objective collaborative optimization with a closed-loop self-learning calibration mechanism and adapt to the complex dynamic environment of multi-story pig farming. There is an urgent need to develop such innovative technical solutions. Summary of the Invention

[0020] The technical problem to be solved by the present invention is to provide an intelligent scheduling method and system that can achieve coordinated optimization of energy consumption, time and balance of feed delivery in multi-story pig farms and has self-learning calibration capabilities, so as to overcome the defects of high energy consumption, unstable efficiency, uneven distribution and lack of intelligence in the existing technology.

[0021] To solve the above-mentioned technical problems, the present invention provides a method for scheduling feed delivery in multi-story pig farms based on multi-objective optimization, characterized by the following steps:

[0022] Step (1) Obtain preset static parameters and real-time collected dynamic parameters. The static parameters characterize the physical properties of the feed, the physical attributes of the conveying system and the structure of the pigsty. The dynamic parameters characterize the real-time feeding demand, the status of equipment and materials and external environmental factors.

[0023] Step (2) Based on the static and dynamic parameters, establish a multi-objective optimization model with scheduling scheme as the variable. The optimization objective of the model includes the coordinated optimization of the energy consumption model E(x), the time model T(x), and the transmission balance model U(x). The energy consumption model E(x) calculates the estimated total energy consumption, the time model T(x) calculates the estimated total time consumption, and the transmission balance model U(x) calculates the estimated transmission balance.

[0024] Step (3) Use a preset optimization algorithm to solve the multi-objective optimization model, obtain the optimal scheduling scheme and generate equipment control instructions;

[0025] Step (4) During or after the execution of the optimal scheduling scheme, a closed-loop self-learning calibration mechanism is initiated, the mechanism including:

[0026] Step (4a) Measure the actual energy consumption data of at least one device during the execution of the optimal scheduling scheme, and calculate the actual total energy consumption accordingly;

[0027] Step (4b) compares the actual total energy consumption with the predicted total energy consumption based on the multi-objective optimization model to determine the deviation between the two;

[0028] Step (4c): When the deviation meets the preset deviation threshold condition, at least one model parameter related to energy consumption calculation in the multi-objective optimization model is reversed to reduce the deviation between the predicted total energy consumption and the actual total energy consumption in subsequent scheduling.

[0029] In the above technical solution, the multi-objective optimization model aims to collaboratively optimize energy consumption, time, and transmission balance.

[0030] The energy consumption model comprehensively considers the dominant energy consumption in multi-story pig farming, including vertical lifting energy consumption, horizontal conveying energy consumption, and equipment start-up and shutdown energy consumption.

[0031] The time model evaluates overall efficiency by calculating the difference between the earliest start time and the latest end time of parallel tasks.

[0032] The transport balance model quantifies the fairness of the allocation by calculating the coefficient of variation of the demand satisfaction rate at each feeding point.

[0033] One of the core innovations of this invention lies in its closed-loop self-learning calibration mechanism.

[0034] This mechanism can measure the energy consumption data of the scheduling scheme during actual execution and compare it with the model's predictions.

[0035] When the deviation between the two meets the preset conditions, the system will automatically correct the relevant parameters in the energy consumption model (such as the equipment efficiency coefficient), so that the model can continuously approach the real working conditions, thus solving the technical problem of the deviation between the theoretical model and the actual operation.

[0036] In addition, the present invention may also include a dynamic weight adjustment mechanism, which can adjust the relative importance of the three optimization objectives in the solution process based on dynamic factors such as real-time electricity price and material inventory.

[0037] When using genetic algorithms or similar optimization algorithms, their operators can be specially designed to comply with process constraints such as path conflict-free operation and anti-cross-material operation.

[0038] The present invention also provides a system for implementing the above method, comprising a sensing and data acquisition layer, a decision and control core layer, and an execution layer, forming a complete three-layer hardware architecture.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] (1) Achieve multi-objective collaborative optimization: find the best balance point among conflicting objectives.

[0041] (2) Significantly improve system performance: energy consumption, time, and balance are all greatly improved.

[0042] (3) Possesses intelligent self-learning ability: It can dynamically correct model parameters and solve the problem of deviation between theory and practice.

[0043] (4) Highly adaptable and innovative: It can adapt to dynamic conditions and meet special process constraints.

[0044] In summary, this invention not only achieves significant technological advancements in energy consumption, efficiency, and balance, but more importantly, through its unique closed-loop self-learning calibration mechanism, it fundamentally solves the problem that static models in existing technologies cannot adapt to dynamic working conditions, providing a novel and adaptive solution for precise control in the field of intelligent aquaculture. Attached Figure Description

[0045] Figure 1 This is an overall flowchart of the method of the present invention.

[0046] Figure 2 This is a schematic diagram of the network topology of the feed conveying system in a multi-story pig farm, where: P1-1 to P6-4 represent the feeding points on each floor; H1-H6 represent the horizontal conveyor belts on each floor.

[0047] Figure 3 A schematic diagram illustrating the objective function of a multi-objective optimization model.

[0048] Figure 4 The following is a hardware architecture diagram of the system of the present invention, wherein: 10-sensing and data acquisition layer; 20-decision and control core layer; 30-execution layer; 11-hopper sensor; 12-equipment status sensor; 13-data gateway; 21-intelligent scheduling server; 22-communication interface module; 23-database storage; 24-closed-loop self-learning calibration module; 31-execution control unit; 32-speed control drive unit; 33-path switching actuator; 34-conveying equipment. Detailed Implementation

[0049] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand the present invention.

[0050] Reference Figure 4 The present invention provides a scheduling system. The system physically includes a sensing and data acquisition layer (10), a decision and control core layer (20), and an execution layer (30).

[0051] The core of the decision-making and control core layer (20) is the intelligent scheduling server (21).

[0052] The intelligent scheduling server (21) includes a processor and a memory in hardware. The memory stores computer program instructions, which, when executed by the processor, implement all the steps of the method described in this invention.

[0053] Example 1

[0054] Experimental environment setup:

[0055] Test scenario: A 6-story pig farm with 4 feeding points on each floor, for a total of 24 feeding points;

[0056] Equipment configuration: 2 elevators (15kW, 12kW), 6 horizontal conveyor belts;

[0057] Test period: 30 consecutive days, 6 times a day;

[0058] Comparison benchmark: Traditional fixed-sequence scheduling method, which strictly follows the order of floors from high to low for sequential feeding.

[0059] Step (1): Parameter acquisition.

[0060] In this embodiment, the static parameters specifically include: the unit density ρ and adhesion coefficient μ of the feed; the length L of the conveying path, the vertical lifting height H, and the rated power P of the equipment; and the basic demand D of each feeding point. The dynamic parameters specifically include: the current real-time demand of each feeding point, the real-time remaining feed in the silo, the equipment operating status (e.g., idle, running, faulty), and the real-time time-of-use electricity price.

[0061] Step (2): Establish a multi-objective optimization model.

[0062] Based on the parameters, a multi-objective optimization model with scheduling schemes as variables is established.

[0063] The model is composed of a weighted sum of three sub-models:

[0064] 1. The energy consumption model E(x) reflects the unique characteristics of feed transportation in multi-story pig farming, and its calculation formula is as follows:

[0065]

[0066] In the formula, For scheduling scheme variables, For a set of vertical delivery tasks, For equipment collection, Let be the feed mass (kg) for the i-th task. Let gravitational acceleration be 9.8 m / s². Let be the vertical lifting height (m) of the i-th task. To improve the mechanical efficiency of the machine (dimensionless). The horizontal power (kW) is supplied for the i-th task. Let the horizontal transport distance (m) be the i-th task. Let be the delivery speed (m / s) of the i-th task. Let j be the number of times the j-th device is started. Let be the energy consumption (kWh) of a single startup of the j-th device.

[0067] In this model, vertical lifting energy consumption is dominant, which is a key feature that distinguishes multi-story pig farming from traditional planar farming. The model comprehensively considers three parts: vertical lifting energy consumption, horizontal conveying energy consumption, and equipment start-up and shutdown energy consumption.

[0068] 2. The time model T(x) is calculated based on the time span of parallel tasks, and its calculation formula is as follows:

[0069]

[0070] In the formula, For scheduling scheme variables, This is the set of end times for all tasks. This is the set of start times for all tasks. This is the latest time the task can end. This is the earliest task start time.

[0071] This model can effectively evaluate the overall efficiency of parallel scheduling, avoiding the overestimation problem that may be caused by simply accumulating the time of each task, and is more in line with the characteristics of multi-device collaborative work in actual production.

[0072] 3. The transport balance model U(x) uses the coefficient of variation of the demand satisfaction rate to quantify it, and its calculation formula is as follows:

[0073]

[0074] In the formula, For scheduling scheme variables, This is the coefficient of variation. Let the demand satisfaction rate be the rate at the k-th feeding point. The standard deviation of the demand satisfaction rate at each feeding point. This represents the average demand satisfaction rate at each feeding point.

[0075] This model uses statistical methods to quantify the balance of feed distribution among different feeding points, which is directly related to the uniformity of pig growth and breeding efficiency.

[0076] The final collaborative optimization objective function is:

[0077]

[0078] In the formula, For scheduling scheme variables, This is the normalized energy consumption value. The normalized time value. To transmit the balance value, Energy consumption weighting coefficient For time weighting coefficients, This is the weighting coefficient for balance.

[0079] Step (3): Solve the model.

[0080] This step uses a preset optimization algorithm to solve the model. In a preferred embodiment, a genetic algorithm is used.

[0081] Of course, particle swarm optimization, simulated annealing, and other algorithms can also be selected, all of which fall within the protection scope of this invention.

[0082] In this embodiment, the crossover and mutation operators of the genetic algorithm are specially designed to comply with path conflict-free and cross-contamination prevention constraints.

[0083] Customized genetic operator implementation:

[0084] 1. Task block crossover operator: Randomly select the crossover point, swap the parent task blocks, and then perform a validity check. If there is a path conflict, adjust the time window of the conflicting task to ensure that a valid child is generated.

[0085] 2. Anti-cross-feed mutation operator: Randomly selects the mutation location, checks for feed type conflicts, and automatically inserts a 3-minute cleanup interval if different feed types overlap in time windows on the same path.

[0086] Step (4): Closed-loop self-learning calibration.

[0087] (4a) Measure relevant data of actual energy consumption: Collect current and voltage data of the main energy-consuming equipment by using the current transformer installed in the motor control cabinet of the main energy-consuming equipment, and calculate the actual total energy consumption.

[0088] (4b) Compare the deviation: Compare the deviation between the actual total energy consumption and the model-predicted total energy consumption.

[0089] (4c) Model Correction: When the deviation meets the preset deviation threshold condition (e.g., the absolute deviation exceeds 5% of the predicted value, or three consecutive deviations are in the same direction), the gradient descent method or other algorithms are used to correct at least one model parameter related to energy consumption calculation (e.g., hoist mechanical efficiency) in the model. wait).

[0090] Specific implementation of gradient descent:

[0091] 1. Define the error function:

[0092] In the formula, Let η be the error function with the efficiency parameter as the variable. The total energy consumption (kWh) is actually measured using devices such as current sensors. Based on the current efficiency parameters The predicted total energy consumption (kWh) calculated by the energy consumption model. For equipment efficiency parameters (such as the mechanical efficiency of a hoist) (e.g., horizontal conveying efficiency), with values ​​typically ranging from 0.6 to 0.95.

[0093] 2. Calculate the partial derivatives:

[0094] In the formula, For the error function with respect to the efficiency parameter The partial derivative of represents the direction of the gradient of the error function. This represents the deviation (kWh) between actual and predicted energy consumption. To predict the effect of the energy consumption function on the efficiency parameter The partial derivatives of .

[0095] 3. Parameter update:

[0096] In the formula, The updated efficiency parameter values. This represents the efficiency parameter value in the current iteration. The learning rate controls the step size of parameter updates, typically ranging from 0.001 to 0.01. A value that is too large may cause oscillations, while a value that is too small will slow down the convergence speed.

[0097] 4. Iterative convergence: Repeat the above process until the error function is minimized.

[0098] Convergence conditions include: the error change between two consecutive iterations is less than a threshold, or the maximum number of iterations (e.g., 50) is reached, or the gradient value is close to zero.

[0099] To verify the beneficial effects of the present invention, the method of the present invention was compared with the traditional scheduling method in the above-mentioned experimental environment.

[0100] Table 1: Performance Comparison of the Invention Method and Traditional Scheduling Methods

[0101] Comparison items Traditional scheduling methods The method of this invention (energy saving priority) The method of this invention (efficiency first) Total energy consumption (kWh) 150 125(-16.7%) 138(-8.0%) Total time (hours) 2.5 2.1(-16.0%) 1.7(-32.0%) Balance (standard deviation) 0.15 0.04(-73.3%) 0.06(-60.0%)

[0102] In addition, the effectiveness of the self-learning mechanism was tracked and simulated.

[0103] In the early stages of system operation, the prediction error rate of the energy consumption model fluctuated within ±8%.

[0104] After about 30 scheduling cycles of self-learning and calibration, the model's prediction bias rate converged from the initial ±8% and stabilized within ±2%, while the prediction accuracy improved from 85% to over 96%, demonstrating the effectiveness of the mechanism.

[0105] In practical applications, when the real-time electricity price is detected to be at its peak, the system automatically adjusts the weights to α=0.6 (increasing the energy consumption weight), β=0.25, and γ=0.15, prioritizing energy conservation; when the feed warehouse inventory is detected to be below 20%, the system further increases the balance weight γ to ensure that all feeding points can obtain basic feed supply.

[0106] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention.

[0107] All equivalent structural or procedural transformations made based on the description and drawings of this invention, or direct or indirect applications to other related technical fields, are similarly included within the scope of patent protection of this invention.

Claims

1. A method for scheduling feed delivery in multi-story pig farms based on multi-objective optimization, characterized in that, Includes the following steps: (1) Obtain preset static parameters and real-time collected dynamic parameters. The static parameters represent the physical properties of feed, the physical properties of the conveying system and the structure of the pig house. The dynamic parameters represent the real-time feeding demand, the status of equipment and materials and external environmental factors. (2) Based on the static and dynamic parameters, a multi-objective optimization model with scheduling scheme as the variable is established. The optimization objective of the model includes the coordinated optimization of the energy consumption model E(x), the time model T(x), and the transmission balance model U(x). The energy consumption model E(x) calculates the estimated total energy consumption, the time model T(x) calculates the estimated total time consumption, and the transmission balance model U(x) calculates the estimated transmission balance. (3) Solve the multi-objective optimization model using a preset optimization algorithm to obtain the optimal scheduling scheme and generate equipment control instructions; (4) During or after the execution of the optimal scheduling scheme, a closed-loop self-learning calibration mechanism is initiated, the mechanism including: (4a) Measure the actual energy consumption data of at least one device during the execution of the optimal scheduling scheme, and calculate the actual total energy consumption accordingly; (4b) Compare the actual total energy consumption with the predicted total energy consumption based on the multi-objective optimization model to determine the deviation between the two; (4c) When the deviation meets the preset deviation threshold condition, at least one model parameter related to energy consumption calculation in the multi-objective optimization model is reversed to reduce the deviation between the predicted total energy consumption and the actual total energy consumption in subsequent scheduling.

2. The method according to claim 1, characterized in that, In step (1), the static parameters include the unit density and adhesion coefficient of the feed; the equipment parameters include the length of the conveying path, the vertical lifting height and the rated power of the equipment; and the pig house parameters include the demand of each feeding point.

3. The method according to claim 1 or 2, characterized in that, The energy consumption model E(x) is established to take into account the energy consumption difference between vertical lifting and / or horizontal conveying, and to include the energy consumption of equipment start-up and shutdown.

4. The method according to claim 3, characterized in that, The energy consumption model E(x) comprehensively considers three components: vertical lifting energy consumption, horizontal conveying energy consumption, and equipment start-up and shutdown energy consumption. Vertical lifting energy consumption is calculated based on conveying volume, lifting height, and equipment efficiency; horizontal conveying energy consumption is calculated based on equipment power, conveying distance, and conveying speed; and equipment start-up and shutdown energy consumption is calculated based on the number of equipment starts and the energy consumption per start.

5. The method according to any one of claims 1-4, characterized in that, The time model T(x) is determined by calculating the difference between the earliest start time and the latest end time of all tasks in the scheduling scheme.

6. The method according to any one of claims 1-5, characterized in that, The delivery balance model U(x) is determined by calculating the coefficient of variation of the demand satisfaction rate at each feeding point.

7. The method according to any one of claims 1-6, characterized in that, In step (4c), gradient descent or a variant thereof is used to correct the model parameters in reverse.

8. The method according to any one of claims 1-7, characterized in that, The optimization algorithm is a genetic algorithm, and its crossover and / or mutation operators are designed to comply with preset path conflict-free and / or anti-cross-material constraints when generating offspring scheduling schemes.

9. The method according to any one of claims 1-8, characterized in that, The method also includes a dynamic adjustment step, which adjusts the weights of energy consumption E(x), time consumption T(x), and balance degree U(x) in the collaborative optimization based on real-time electricity price and warehouse inventory status.

10. A multi-story pig farm feed conveying and scheduling system based on multi-objective optimization, characterized in that, include: (1) Sensing and data acquisition layer, including silo sensors, equipment status sensors and data gateway; (2) The decision-making and control core layer includes an intelligent scheduling server on which a multi-objective optimization engine runs, the engine including: The data management module is used to obtain static and dynamic parameters; The collaborative optimization module is used to establish and solve a multi-objective optimization model that collaboratively optimizes energy consumption, time, and balance. A closed-loop self-learning calibration module is configured to iteratively correct the efficiency parameters in the energy consumption model by comparing the predicted energy consumption of the scheduling scheme with the actual measured energy consumption after execution. The scheduling plan generation module is used to generate equipment control instructions based on the optimization results; (3) Execution layer, including execution control unit, speed control drive unit, and path switching actuator; The system is configured to perform the method of any one of claims 1-9.

11. The system according to claim 10, characterized in that, The closed-loop self-learning calibration module further includes: (1) Actual energy consumption data acquisition unit, used to acquire the actual power of the equipment through a current sensor and calculate the actual total energy consumption; (2) Deviation analysis unit, used to calculate the systematic deviation between predicted energy consumption and actual energy consumption; (3) Parameter optimization unit, used to trigger the gradient descent algorithm to correct efficiency parameters when the deviation exceeds the threshold; (4) Model update unit, used to save the corrected parameters and apply them to subsequent scheduling calculations.

12. The system according to claim 10, characterized in that, The silo sensor is a weighing sensor or radar level gauge installed at the bottom of the feed silo to monitor the amount of remaining material in real time; the equipment status sensor includes a current transformer integrated in the motor control cabinet and an ultrasonic or infrared blockage sensor installed at key nodes of the pipeline; the execution control unit is a PLC or industrial PC, the speed control drive unit is a frequency converter, and the path switching actuator is an electric valve or a three-way distributor.

Images