Method, system and computing device for intelligent farming based on soft bus

Abstract

This application provides a method, system, and computing device for intelligent aquaculture based on a soft bus. The method is applied to a mobile feeding device equipped with a soft bus communication module. The method includes: before entering the target shed, automatically establishing a second near-field network with the target feeder within the near-field range of the feed warehouse via the soft bus communication module; sending a feeding control command to the target feeder through the second near-field network to cause the target feeder to add feed to the mobile feeding device; after feed addition, automatically establishing a first near-field network with sensing devices inside the target shed within the near-field range of the target shed via the soft bus communication module; acquiring data collected by the sensing devices through the first near-field network; determining the target feeding strategy for the target shed based on the data and a preset feeding plan; and executing the feed delivery task according to the target feeding strategy. This solution can automate and intelligentize the feeding process in a livestock farm.

Description

Technical Field

[0001] This application relates to the field of animal husbandry, and more specifically, to a method, system, and computing device for intelligent animal husbandry based on a soft bus. Background Technology

[0002] In large-scale livestock farming, with the advancement of smart farm construction, the feeding process has evolved from simple mechanical operation to an information-based operation process involving multiple devices. This generally involves the following key equipment: monitoring cameras, environmental sensors, feeders, mobile feeding equipment, etc. These devices can achieve closed-loop management of the entire chain from "formula issuance - raw material loading - mixing - precise feeding - effect monitoring", which is a key technology link for reducing costs and increasing efficiency in large-scale farms.

[0003] In one related feed feeding technical solution, surveillance cameras and environmental sensors are used to monitor the real-time status of the livestock sheds. Based on this real-time information, the feed formulation, loading, and feeding strategies are manually adjusted. Corresponding loading and feeding instructions are then issued to drive the loading machine and mobile feeding equipment to perform feed loading and feeding actions. However, this solution suffers from high labor costs and low efficiency. Furthermore, because monitoring equipment, sensing devices, and operating equipment are often provided by different manufacturers, and communication protocols, interface specifications, and data formats are inconsistent, system integration is difficult. The lack of a unified interoperability mechanism between different devices further restricts the automation and intelligence level of the feeding process.

[0004] Therefore, how to automate and intelligentize the feeding process in livestock farms has become an urgent technical problem to be solved. Summary of the Invention

[0005] This application provides a method, system, and computing device for intelligent aquaculture based on a soft bus, which can realize the automation and intelligence of the feeding process in aquaculture farms.

[0006] Firstly, a method for intelligent aquaculture based on a soft bus is provided. This method is applied to a mobile feeding device equipped with a soft bus communication module. The method includes: before the mobile feeding device enters the target shed, within the near-field range of the feed warehouse, automatically establishing a second near-field network with a target feeder via the soft bus communication module, the target feeder also equipped with a soft bus communication module; sending a feeding control command to the target feeder via the second near-field network, causing the target feeder to add feed to the mobile feeding device according to the feeding list; after the mobile feeding device completes feeding, within the near-field range of the target shed, automatically establishing a first near-field network with a sensing device deployed within the target shed via the soft bus communication module, the sensing device also equipped with a soft bus communication module; acquiring on-site sensing data collected by the sensing device via the first near-field network; determining a target feeding strategy for the target shed based on the on-site sensing data and a preset feeding plan; and executing the feeding task according to the target feeding strategy.

[0007] In the aforementioned technical solution, the mobile feeding device, acting as a cross-regional mobile control node, leverages the automatic discovery and dynamic networking capabilities of its soft bus communication module to achieve cross-physical-space roaming dynamic networking and end-to-end automated collaboration. Specifically, before entering the target shed, the mobile feeding device can automatically coordinate with the target loading machine in the feed warehouse to complete the quantitative loading of upstream feed. Subsequently, upon entering the target shed after loading, it can automatically connect to downstream sensing devices and aggregate on-site sensing data, thereby dynamically adjusting and executing a precise feeding strategy. This solution not only reduces the integration difficulty of equipment from different manufacturers but also completely establishes an end-to-end business loop from "upstream raw material loading to order" to "downstream environmental status sensing and intelligent feeding," breaking through the technical limitations of fixed equipment and single-point control in existing technologies. This significantly improves the continuity of overall feed preparation and on-demand supply, the adaptability of strategies in abnormal scenarios, and the overall intelligent operation efficiency of the entire system.

[0008] In conjunction with the first aspect, in some implementations of the first aspect, the target feeding strategy includes at least one of the following: target single feeding amount, target feeding frequency, target discharge speed, and target execution action. In the above technical solution, by clearly defining quantitative control parameters such as single feeding amount, frequency, discharge speed and execution action in the strategy, precise operating indicators and operation direction guidance are provided for the underlying electromechanical actuator of the mobile feeding equipment, realizing the standardization and precise quantification of feeding execution actions.

[0009] In conjunction with the first aspect, in some implementations of the first aspect, an input dataset is determined based on the field perception data and the feeding plan, wherein the input dataset includes the planned single feeding amount, the planned feeding frequency, the field status data of the target shed, and the remaining amount of material in the mobile feeding device; and the target feeding strategy is determined based on the input dataset. In the above technical solution, a multi-dimensional input dataset is constructed by integrating planned task parameters, environmental dynamic variables, and the physical load boundaries of the equipment itself. This provides a data foundation with a global perspective for the strategy generation module, avoiding the risk of decision-making bias or physical overreach caused by a single data source from the source.

[0010] In conjunction with the first aspect, in some implementations of the first aspect, the target feeding strategy is calculated based on the input dataset and the preset feeding strategy formula. In the above technical solution, the system directly performs logical calculations using a preset feeding strategy formula, which gives the system decision-making capabilities with low computing power consumption, strong result certainty, and fast response speed, thus establishing a high-safety basic control baseline for the entire feeding operation.

[0011] In conjunction with the first aspect, in some implementations of the first aspect, the planned single feeding amount and / or planned feeding frequency are corrected using correction coefficients determined based on the field status data of the target shed, to obtain the target single feeding amount and target feeding frequency; the target discharge rate is calculated based on the target single feeding amount and the planned feeding time in the feeding plan. In the above technical solution, the standard planned amount and frequency are intervened in real time by using the correction coefficient based on on-site data. This can quickly respond to sudden situations such as livestock’s immediate loss of appetite, feed trough accumulation or physical passage blockage, and achieve dynamic and precise replenishment or reduction of feed, effectively preventing feed accumulation, deterioration and waste.

[0012] In conjunction with the first aspect, in some implementations of the first aspect, the target feeding strategy is obtained by calling an artificial intelligence (AI) big model based on the input dataset, wherein the input information of the AI ​​big model includes the input dataset, and the output information of the AI ​​big model includes the target feeding strategy. In the above technical solution, by calling AI large models to perform deep semantic reasoning and multimodal feature analysis, it is possible to accurately capture the implicit behavioral trends of livestock groups caused by micro-environmental stress that are difficult to quantify by traditional fixed formulas, thereby significantly improving the foresight and adaptability of feeding strategies in long-tail abnormal scenarios.

[0013] In conjunction with the first aspect, in some implementations of the first aspect, a first feeding strategy is calculated based on the input dataset and a preset feeding strategy formula; a second feeding strategy is obtained by calling the AI ​​large model based on the input dataset; and the first feeding strategy and the second feeding strategy are fused to obtain the target feeding strategy. The above technical solution adopts a decision-making logic that combines rule formulas and AI models in parallel. This not only leverages the high intelligence and flexibility of the large model, but also preserves the safety baseline of the physical calculation formula. This effectively prevents the large AI model from generating control "illusions" in extreme scenarios and ensures the robustness of the overall near-field linkage system.

[0014] In conjunction with the first aspect, in some implementations of the first aspect, the deviation value between the first feeding strategy and the second feeding strategy in terms of control parameters and / or execution actions is calculated; based on the deviation value, the target feeding strategy is determined using the corresponding fusion rule.

[0015] In the above technical solution, by accurately quantifying the specific deviations of the two strategies at the underlying parameters and macro-action levels, an objective and clear mathematical basis is provided for the system to automatically determine the reliability limits of AI suggestions and implement a differentiated strategy fusion mechanism.

[0016] In conjunction with the first aspect, in some implementations of the first aspect, when the deviation value is less than a preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent, the second feeding strategy is determined as the target feeding strategy; or when the deviation value is greater than or equal to the preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent, the control parameters in the first feeding strategy and the second feeding strategy are weighted and fused to obtain the target feeding strategy; or when the execution actions of the first feeding strategy and the second feeding strategy are inconsistent, an alarm is triggered to obtain the target feeding strategy for manual confirmation.

[0017] The above technical solution constructs a hierarchical response strategy fusion mechanism, which seamlessly absorbs the benefits of AI fine-tuning under normal working conditions, incorporates the weight of AI's forward-looking predictions within a safe range, and forcibly switches to human arbitration when serious conflicts occur in core execution actions, thus constructing a tight protection closed loop that takes into account both the efficiency of AI decision-making and the bottom line of operational safety.

[0018] In conjunction with the first aspect, in some implementations of the first aspect, the sensing device includes a surveillance camera and an environmental sensor, and the on-site sensing data includes at least one of the following: temperature or humidity collected by the environmental sensor, images of feed troughs collected by the surveillance camera, images of livestock feeding, and images of shed passageways. In the above technical solution, visual cameras and environmental sensors are used together to obtain the multimodal status of the target shed, providing direct and real on-site objective perception input for assessing the degree of leftover material accumulation, livestock feeding desire, climate stress factors and physical driving obstacles.

[0019] Secondly, a device for intelligent aquaculture based on a soft bus is provided. This device is deployed in a mobile feeding device, which also includes a soft bus communication module. The device comprises a creation module, an acquisition module, a determination module, and an execution module. The creation module automatically establishes a second near-field network with the target feeder within the near-field range of the feed warehouse before the mobile feeding device enters the target shed. The target feeder also has a soft bus communication module deployed in it. The acquisition module sends feeding control commands to the target feeder through the second near-field network, causing the target feeder to... The feed loading list adds feed to the mobile feeding device; the creation module is also used to automatically establish a first near-field network with the sensing device deployed in the target shed via the soft bus communication module after the mobile feeding device completes feed loading, within the near-field range of the target shed; the sensing device is equipped with a soft bus communication module; the acquisition module is also used to acquire the field sensing data collected by the sensing device through the first near-field network; the determination module is used to determine the target feeding strategy for the target shed based on the field sensing data and the preset feeding plan; the execution module is used to execute the feed feeding task according to the target feeding strategy.

[0020] In conjunction with the second aspect, in some implementations of the second aspect, the target feeding strategy includes at least one of the following: target single feeding amount, target feeding frequency, target discharge speed, and target execution action.

[0021] In conjunction with the second aspect, in some implementations of the second aspect, the determining module is specifically used to: determine an input dataset based on the field perception data and the feeding plan, wherein the input dataset includes the planned single feeding amount, the planned feeding frequency, the field status data of the target shed, and the remaining amount of material in the mobile feeding device; and determine the target feeding strategy based on the input dataset.

[0022] In conjunction with the second aspect, in some implementations of the second aspect, the determining module is specifically used to: calculate the target feeding strategy based on the input dataset and the preset feeding strategy formula.

[0023] In conjunction with the second aspect, in some implementations of the second aspect, the determining module is specifically used to: use correction coefficients determined based on the on-site status data of the target shed to correct the planned single feeding amount and / or the planned feeding frequency to obtain the target single feeding amount and the target feeding frequency; and calculate the target discharge rate based on the target single feeding amount and the planned feeding time in the feeding plan.

[0024] In conjunction with the second aspect, in some implementations of the second aspect, the determining module is specifically used to: call an artificial intelligence (AI) big model based on the input dataset to obtain the target feeding strategy, wherein the input information of the AI ​​big model includes the input dataset, and the output information of the AI ​​big model includes the target feeding strategy.

[0025] In conjunction with the second aspect, in some implementations of the second aspect, the determining module is specifically used to: calculate a first feeding strategy based on the input dataset and a preset feeding strategy formula; call an AI large model based on the input dataset to obtain a second feeding strategy; and perform strategy fusion on the first feeding strategy and the second feeding strategy to obtain the target feeding strategy.

[0026] In conjunction with the second aspect, in some implementations of the second aspect, the determining module is specifically used to: calculate the deviation value between the first feeding strategy and the second feeding strategy in terms of control parameters and / or execution actions; and determine the target feeding strategy based on the deviation value using the corresponding fusion rule.

[0027] In conjunction with the second aspect, in some implementations of the second aspect, the determining module is specifically used to: determine the second feeding strategy as the target feeding strategy when the deviation value is less than a preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent; or when the deviation value is greater than or equal to the preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent, perform weighted fusion of the control parameters in the first feeding strategy and the second feeding strategy to obtain the target feeding strategy; or when the execution actions of the first feeding strategy and the second feeding strategy are inconsistent, trigger an alarm prompt to obtain the target feeding strategy for manual confirmation.

[0028] In conjunction with the second aspect, in some implementations of the second aspect, the sensing device includes a surveillance camera and an environmental sensor, and the on-site sensing data includes at least one of the following: temperature or humidity collected by the environmental sensor, images of feed troughs collected by the surveillance camera, images of livestock feeding, and images of shed passageways. It should be understood that for the beneficial effects of the second aspect and its various implementations, please refer to the first aspect and its various implementations; they will not be repeated here.

[0029] Thirdly, a soft bus-based intelligent aquaculture system is provided. This system includes a mobile feeding device, a target feeder deployed within a feed warehouse, and sensing devices deployed within a target shed. Each of the mobile feeding device, the target feeder, and the sensing devices is equipped with a soft bus communication module. Before entering the target shed, the mobile feeding device automatically establishes a second near-field network with the target feeder within the near-field range of the feed warehouse via the soft bus communication module, and sends a feeding control command to the target feeder through this second near-field network. The target feeder is used to respond to the feeding control command. The feed control command adds feed to the mobile feeding device according to the feeding list. After feed addition is completed, the mobile feeding device automatically establishes a first near-field network with the sensing device via the soft bus communication module within the near-field range of the target shed. The sensing device collects on-site sensing data and sends the on-site sensing data to the mobile feeding device via the first near-field network. The mobile feeding device also determines a target feeding strategy for the target shed based on the on-site sensing data and a preset feeding plan. The mobile feeding device also executes the feed delivery task according to the target feeding strategy.

[0030] In conjunction with the third aspect, in some implementations of the third aspect, the target feeding strategy includes at least one of the following: target single feeding amount, target feeding frequency, target discharge speed, and target execution action.

[0031] In conjunction with the third aspect, in some implementations of the third aspect, the mobile feeding device is specifically used to: determine an input dataset based on the field perception data and the feeding plan, wherein the input dataset includes the planned single feeding amount, the planned feeding frequency, the field status data of the target shed, and the remaining amount of material of the mobile feeding device; and determine the target feeding strategy based on the input dataset.

[0032] In conjunction with the third aspect, in some implementations of the third aspect, the mobile feeding device is specifically used to: calculate the target feeding strategy based on the input dataset and the preset feeding strategy formula.

[0033] In conjunction with the third aspect, in some implementations of the third aspect, the mobile feeding device is specifically used to: correct the planned single feeding amount and / or the planned feeding frequency using a correction coefficient determined based on the on-site status data of the target shed, to obtain the target single feeding amount and the target feeding frequency; and calculate the target discharge speed based on the target single feeding amount and the planned feeding time in the feeding plan.

[0034] In conjunction with the third aspect, in some implementations of the third aspect, the mobile feeding device is specifically used to: call the AI ​​big model based on the input dataset to obtain the target feeding strategy, wherein the input information of the AI ​​big model includes the input dataset, and the output information of the AI ​​big model includes the target feeding strategy.

[0035] In conjunction with the third aspect, in some implementations of the third aspect, the mobile feeding device is specifically used to: calculate a first feeding strategy based on the input dataset and a preset feeding strategy formula; call the AI ​​large model based on the input dataset to obtain a second feeding strategy; and perform strategy fusion on the first feeding strategy and the second feeding strategy to obtain the target feeding strategy.

[0036] In conjunction with the third aspect, in some implementations of the third aspect, the mobile feeding device is specifically used to: calculate the deviation value between the first feeding strategy and the second feeding strategy in terms of control parameters and / or execution actions; and determine the target feeding strategy based on the deviation value using the corresponding fusion rule.

[0037] In conjunction with the third aspect, in some implementations of the third aspect, the mobile feeding device is specifically used to: determine the second feeding strategy as the target feeding strategy when the deviation value is less than a preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent; or when the deviation value is greater than or equal to the preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent, perform weighted fusion of the control parameters in the first feeding strategy and the second feeding strategy to obtain the target feeding strategy; or when the execution actions of the first feeding strategy and the second feeding strategy are inconsistent, trigger an alarm prompt to obtain the target feeding strategy for manual confirmation.

[0038] In conjunction with the third aspect, in some implementations of the third aspect, the sensing device includes a surveillance camera and an environmental sensor, and the on-site sensing data includes at least one of the following: temperature or humidity collected by the environmental sensor, images of feed troughs collected by the surveillance camera, images of livestock feeding, and images of shed passageways.

[0039] It should be understood that for the beneficial effects of the third aspect and its various implementations, please refer to the first aspect and its various implementations; they will not be repeated here.

[0040] Fourthly, a computing device is provided, including a processor and a memory, and optionally, an input / output interface. The processor controls the input / output interface to send and receive information, the memory stores a computer program, and the processor retrieves and runs the computer program from the memory, causing the computing device to execute the methods of the first aspect or any possible implementation thereof.

[0041] Optionally, the processor can be a general-purpose processor, which can be implemented in hardware or software. When implemented in hardware, the processor can be a logic circuit, integrated circuit, etc.; when implemented in software, the processor can be a general-purpose processor that reads software code stored in memory. This memory can be integrated into the processor or located outside the processor and exist independently.

[0042] Fifthly, a chip is provided that acquires and executes instructions to implement the methods described in the first aspect and any implementation thereof.

[0043] Optionally, as one implementation, the chip includes a processor and a data interface, through which the processor reads instructions stored in the memory and executes the methods in the first aspect and any implementation thereof.

[0044] Optionally, as one implementation, the chip may further include a memory storing instructions, and the processor is used to execute the instructions stored in the memory. When the instructions are executed, the processor is used to perform the method in the first aspect and any implementation thereof.

[0045] In a sixth aspect, a computer program product containing instructions is provided, which, when executed by a computing device, cause the computing device to perform the methods described in the first aspect and any implementation thereof.

[0046] In a seventh aspect, a computer-readable storage medium is provided, including computer program instructions that, when executed by a computing device, perform the method as described in the first aspect and any implementation thereof.

[0047] As examples, these computer-readable storage devices include, but are not limited to, one or more of the following: read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), flash memory, electrically EPROM (EEPROM), and hard drive.

[0048] Alternatively, as one implementation method, the aforementioned storage medium can specifically be a non-volatile storage medium. Attached Figure Description

[0049] Figure 1 This is a schematic block diagram illustrating the application scenarios applicable to this application.

[0050] Figure 2 This is a schematic flowchart illustrating an intelligent aquaculture method based on a soft bus, as provided in an embodiment of this application.

[0051] Figure 3 This is a schematic flowchart illustrating a method for a mobile feeding device to obtain feed to be fed from a feed warehouse, as provided in an embodiment of this application.

[0052] Figure 4 This is a schematic block diagram illustrating data synchronization between devices via a soft bus communication module, as provided in an embodiment of this application.

[0053] Figure 5 This is a schematic flowchart illustrating a method for a mobile feeding device to determine a target feeding strategy using preset rule formulas and an AI large model, as provided in an embodiment of this application.

[0054] Figure 6 This is a schematic block diagram of an intelligent aquaculture device 600 based on a soft bus, provided in an embodiment of this application.

[0055] Figure 7 This is a schematic diagram of the architecture of a computing device 700 provided in an embodiment of this application. Detailed Implementation

[0056] The technical solutions in this application will now be described with reference to the accompanying drawings.

[0057] This application will present various aspects, embodiments, or features relating to systems comprising multiple devices, components, modules, etc. It should be understood and appreciated that individual systems may include additional devices, components, modules, etc., and / or may not include all devices, components, modules, etc. discussed in conjunction with the accompanying drawings. Furthermore, combinations of these approaches are also possible.

[0058] Furthermore, in the embodiments of this application, the words "exemplary," "for example," etc., are used to indicate that they are examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" in this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the term "exemplary" is intended to present the concept in a concrete manner.

[0059] In the embodiments of this application, "corresponding" and "corresponding" can sometimes be used interchangeably. It should be noted that when the distinction is not emphasized, their intended meanings are consistent.

[0060] The business scenarios described in the embodiments of this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided in the embodiments of this application. As those skilled in the art will know, with the evolution of network architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0061] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0062] In this application, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A alone, A and B simultaneously, and B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0063] In large-scale livestock farming, total mixed ration (TMR) feeding technology has become the mainstream feed administration method. The core of this technology lies in the precise proportioning of roughage, concentrate, and various nutritional additives (such as vitamin and mineral premixes) according to a scientific formula based on the nutritional needs of livestock and poultry at different growth stages. After thorough mixing, these are fed in a uniform form in a single feeding, thereby avoiding picky eating by animals, ensuring balanced nutrient intake, and significantly improving feed conversion rate and farming efficiency.

[0064] With the advancement of smart farm construction, the TMR feeding process has evolved from simple mechanical operation to an information-based operation process involving multiple devices. It generally involves the following key equipment: monitoring cameras, environmental sensors, feeders, mobile feeding equipment, etc. These devices can achieve closed-loop management of the entire chain from "formula issuance - raw material loading - mixing - precise feeding - effect monitoring", which is a key technology link for reducing costs and increasing efficiency in large-scale farms.

[0065] In actual livestock farming scenarios, the conditions in different shed zones often vary significantly. Specifically, the number of livestock, individual feeding activity, amount of feed remaining in troughs, ambient temperature and humidity, and the presence of abnormal events (such as individual illnesses or equipment malfunctions) differ between shed zones. Under these circumstances, implementing a fixed feeding schedule with uniform loading and dispensing can easily lead to a series of problems. For example, some troughs may experience feed waste due to overfeeding and mold growth, while others may suffer from insufficient feeding, causing livestock to compete for food due to hunger. This further exacerbates the contradiction between feed waste and nutritional mismatch. Therefore, there is an urgent need for an intelligent feeding method that can sense the real-time conditions of each zone and dynamically generate differentiated feeding strategies accordingly to replace the traditional fixed-plan model and truly achieve on-demand feeding and precise supply.

[0066] In a related TMR (Traffic-to-Mortar) feed feeding technical solution, although monitoring cameras and environmental sensors have been introduced to collect real-time data on the livestock shed's status, subsequent feeding and dispensing decisions still heavily rely on human experience. The specific process is as follows: after reviewing the monitoring footage and environmental data, staff manually determine the feeding needs of each zone, then adjust the feed ratio, loading amount, and feeding plan one by one. Finally, loading and dispensing instructions are issued manually, driving the loading machine and mobile feeding equipment to sequentially perform feed loading and mixing actions. This solution does not form a complete closed-loop coordination from environmental perception, status recognition, loading control to mixing and dispensing, resulting in numerous manual intervention steps, slow response speed, high overall operating costs, and low efficiency. Furthermore, because monitoring equipment, sensing equipment, and operating equipment are usually from different manufacturers, and the communication protocols, interface specifications, and data formats used by each party are different, there is a lack of a unified interoperability mechanism between the devices, making system integration difficult and hindering efficient data flow between stages, further restricting the automation and intelligence level of the TMR feeding process.

[0067] In view of this, the present application provides an intelligent aquaculture method based on a soft bus. This method enables mobile feeding equipment to automatically connect to the feeder, camera and environmental sensors in the current area during mobile operation, realize real-time near-field linkage, improve the automation and intelligence level of the feeding process in the farm, and reduce the difficulty of integrating equipment from different manufacturers.

[0068] For ease of description, the following will be combined with... Figure 1 The application scenarios applicable to this application are described in detail.

[0069] Figure 1 This is a schematic block diagram illustrating an application scenario applicable to this application. For example... Figure 1 The application scenario mainly consists of two functional areas: a feed warehouse and livestock sheds. The feed warehouse and livestock sheds are closely linked through the autonomous movement of mobile feeding equipment and the real-time transmission of on-site data: the feed warehouse completes the loading and mixing of feed according to the strategy, while the livestock shed adjusts the feeding rhythm based on the real-time status. The two work together to form a complete closed loop from "perception-decision-execution-feedback".

[0070] The equipment included in the feed warehouse and livestock sheds will be described in detail below.

[0071] 1) Feed warehouse: The feed warehouse is the material preparation and processing center for the entire TMR feeding process, undertaking core operations such as raw material storage, feed loading, and mixing. The feed warehouse is usually divided into roughage storage areas, concentrate storage areas, and additive storage areas, with each component of the feed stored separately to avoid interference.

[0072] For example, such as Figure 1 As shown, the feed warehouse can be equipped with at least one loading machine and at least one mobile feeding device.

[0073] In some embodiments, the mobile feeding device may also be referred to as a mixing feeder.

[0074] The aforementioned feed loading machine is mainly responsible for loading feed. It is used to automatically grab feed (such as roughage, concentrate and nutritional additives) from the corresponding feed storage area according to the feeding formula instructions, and complete the quantitative loading according to the set ratio.

[0075] It should be understood that in actual deployment, the operating range of a single feed loader can be flexibly set according to the scale of the ranch and management needs. For example, a single feed loader can be dedicated to a storage area for a specific type (or category) of feed, only responsible for grabbing and loading that component of raw materials, achieving dedicated machine for dedicated feed and no interference between them. Alternatively, a single feed loader can simultaneously cover storage areas for multiple types (or categories) of feed, sequentially completing the quantitative grabbing and loading of different component raw materials on the same machine, thereby reducing the number of devices and lowering hardware investment. That is, the embodiments of this application do not specifically limit the types and quantities of feed handled by the feed loader.

[0076] The aforementioned mobile feeding equipment mainly undertakes the function of mixing and feeding. It is mainly responsible for fully mixing the various feeds loaded by the feeder in its mixing chamber, so that the components are uniform and consistent to form the finished daily ration. It also has the ability to move autonomously, and can move from the feed warehouse to each livestock shed along the planned route to evenly transport the finished daily ration to each livestock shed and perform feeding operations.

[0077] It should be noted that the embodiments of this application do not specifically limit the number of loading machines and mobile feeding devices installed in the feed warehouse. Figure 1 The example described is based on a feed warehouse equipped with two feeders (e.g., feeder 1 and feeder 2) and one mobile feeding device.

[0078] 2) Livestock sheds: The livestock shed is the terminal operation area of ​​the TMR feeding process, responsible for livestock feeding and on-site condition monitoring. Specifically, the livestock shed is functionally divided into feed trough area and environmental monitoring area. The feed trough area is used to receive finished rations delivered by mobile feeding equipment for livestock to eat freely, while the environmental monitoring area is equipped with monitoring cameras and environmental sensors.

[0079] The aforementioned surveillance cameras primarily perform visual detection functions, capturing real-time images of the feeding area to assist in identifying feeding status, abnormal animal behavior, and equipment operation.

[0080] The aforementioned environmental sensors primarily function as environmental sensors, continuously monitoring parameters such as temperature, humidity, and dust concentration in the material tank area.

[0081] The following is combined Figure 2 This application provides a detailed description of an intelligent aquaculture method based on a soft bus, as illustrated in the embodiments of this application. It should be understood that... Figure 2 The examples are merely to help those skilled in the art understand the embodiments of this application, and are not intended to limit the embodiments of the application to... Figure 2 The specific numerical values ​​or specific scenarios illustrated. Those skilled in the art will understand based on... Figure 2 The examples given below can obviously be modified or varied in various ways, and such modifications and variations also fall within the scope of the embodiments of this application.

[0082] Figure 2 This is a schematic flowchart illustrating an intelligent aquaculture method based on a soft bus, as provided in an embodiment of this application. Figure 2 As shown, the method may include steps 210-240, which will be described in detail below.

[0083] Step 210: The mobile feeding device retrieves the feed to be fed from the feed warehouse.

[0084] In this embodiment of the application, the mobile feeding device can enter the feed warehouse and obtain the feed to be fed from the feed warehouse.

[0085] The following is combined Figure 3 The paper describes in detail the specific process of a mobile feeding device entering a feed warehouse and retrieving feed from the warehouse.

[0086] For example, such as Figure 3 As shown, the process of a mobile feeding device obtaining feed to be fed from a feed warehouse may include steps 1-7, which are described in detail below.

[0087] Step 1: The mobile feeding equipment starts and enters the near-field range of the feed warehouse. It discovers the loader via soft bus broadcast and sends a network request.

[0088] For example, such as Figure 1 and Figure 4 As shown, the mobile feeding device, which serves as both a mobile operation node and the control decision-making brain, is equipped with a soft bus communication module. After the mobile feeding device is started, it can enter the near-field range of the feed warehouse according to the planned route.

[0089] It should be understood that the aforementioned soft bus communication module is the core component for achieving efficient cross-device collaboration. Its main function is to break down information barriers between different manufacturers, different hardware capabilities, and underlying communication protocols by building a virtual bus network at the underlying level. Through a unified soft bus communication module acting as a communication bridge, devices with different hardware capabilities and underlying protocols (such as mobile feeding devices, loading machines, surveillance cameras, environmental sensors, etc.) can achieve automatic discovery, connection, dynamic networking, and data transmission within the same local area network. This enables multiple devices within the same local area network to achieve automatic discovery, dynamic networking, and unified data transmission.

[0090] The aforementioned near-field range refers to the intersection of physical distance and the effective coverage area of ​​short-range wireless communication technology. In other words, when the mobile feeding equipment enters this specific space, the two conditions of "sufficient physical distance" and "wireless signal reachability" are simultaneously met, meaning that the physical conditions for establishing short-range wireless communication between the equipment are met.

[0091] For example, the near-field range of the aforementioned feed warehouse may include, but is not limited to, the wireless signal coverage area of ​​Wi-Fi or Bluetooth BT. Within this near-field range, operating equipment such as feeders and mobile feeding devices in the feed warehouse can establish a stable connection through the aforementioned unified soft bus communication module to realize data interaction such as issuing feeding commands, reporting equipment status, and configuring mixing parameters, thereby meeting the short-range, high-efficiency communication needs between devices within the feed warehouse.

[0092] For example, such as Figure 1 and Figure 4 As shown, the aforementioned soft bus communication module is also deployed in the feed loader located in the feed warehouse. When the mobile feeding device enters the near-field range of the feed warehouse, the physical distance between the mobile feeding device and the feed loader is shortened to within the effective coverage radius of short-range wireless communication technologies (such as Wi-Fi or Bluetooth), thus fulfilling the physical conditions for establishing a short-range wireless communication connection. Based on this condition, the mobile feeding device can actively send a network request to the feed loader.

[0093] Specifically, the mobile feeding device can actively broadcast via the soft bus communication module to automatically discover the target feeder located in the feed warehouse area and actively send a network request to the target feeder.

[0094] The aforementioned target loading machine can be one loading machine located in a feed warehouse, or it can be multiple loading machines located in a feed warehouse. This application embodiment does not specifically limit this.

[0095] In the above technical solution, both the target loading machine and the mobile feeding device are equipped with a unified soft bus communication module. Through the unified soft bus communication module as a communication bridge, the target loading machine and the mobile feeding device establish a near-field linkage network between the originally independently operating devices through a "broadcast-response" mechanism that does not require manual intervention. This network is used for issuing control commands and transmitting loading progress.

[0096] Step 2: The target loading machine responds to the networking request of the mobile feeding device and returns status information to the mobile feeding device.

[0097] For example, when a target loader located within the near-field range receives a broadcast from the mobile feeding device and a networking request, the target loader will respond through its internally deployed soft bus communication module to establish a second near-field network.

[0098] Specifically, during the response process, the target feeder can proactively return its current key status information to the mobile feeding device, which acts as the control node, based on the established short-range wireless communication connection (second near-field network).

[0099] The key status information returned by the target feeder may include, but is not limited to: feeder equipment identifier, hopper location information, feed type, and inventory status. The feeder equipment identifier is used to uniquely identify the target feeder entity within the near-field dynamic network, establishing a clear communication address. The hopper location information indicates the precise physical location or coordinates of the target feeder (or its specific hopper), enabling the mobile feeding device to accurately move and dock at the target work point in subsequent steps. The feed type clearly indicates the specific type of feed currently stored by the target feeder (such as a specific type of roughage, concentrate, or nutritional additives), which is crucial for the mobile feeding device to match the TMR total mixed ration formulation. The inventory status provides real-time feedback on the remaining feed quantity in the hopper, ensuring that the node has the physical conditions to meet the expected loading task.

[0100] In the above technical solution, through the network response and multi-dimensional status information feedback, the loading machine and the mobile feeding device successfully completed dynamic networking and underlying data handshake based on the soft bus. At this point, the mobile feeding device has fully grasped the operational resources available at the current node, thus providing the necessary data input for formulating a specific loading execution plan in the next step.

[0101] Step 3: The mobile feeding device generates a loading list based on the feeding formula.

[0102] For example, after obtaining the status information of the target feeder in the near field range (including feeder equipment identification, hopper location information, feed type, inventory status, etc.) through the soft bus communication module, the mobile feeding device can generate a feeding list based on the feeding formula.

[0103] The above-mentioned feeding formula can be preset by the system or input by the user. This application embodiment does not specifically limit this.

[0104] Specifically, the system's preset feeding formula can be a standard formula pre-installed at the factory, or a recommended formula generated during equipment operation based on historical data, empirical parameters, or algorithm optimization. User-defined feeding formulas allow users to flexibly adjust parameters such as the type, proportion, and timing of each component in the formula according to actual working conditions, material characteristics, or process requirements, thereby achieving more precise feeding control. Through these methods, this application ensures both the system's out-of-the-box usability and provides users with ample space for personalized configuration, making it suitable for various application scenarios.

[0105] The above-mentioned loading list may include, but is not limited to: the type of feed to be loaded, the target weight, etc.

[0106] The feed type to be loaded can be a single feed type or a combination of multiple feed types. This application does not specifically limit this.

[0107] For example, the above-mentioned loading list can also be referred to as feeding information, such as... Figure 1 As shown, the mobile feeding device can send the feeding information to the target loader (e.g., loader 1 and loader 2).

[0108] Specifically, taking the type of feed to be loaded as an example, the target weight mentioned above refers to the total weight of that type of feed that needs to be loaded, which is applicable to batch loading scenarios of single-variety feed.

[0109] For example, taking TMR feeding technology as an example, the feed types to be loaded include multiple feed types, and the target weight refers to the weight that each type of feed needs to be loaded to ensure that the proportion of each component of the mixed diet meets the nutritional formula requirements.

[0110] Optionally, taking the case where the feed to be loaded includes multiple types of feed, the above-mentioned loading list may further include the loading order of these multiple types of feed. For example, roughage may be loaded first, followed by concentrate, and finally additives, thereby meeting the differentiated requirements of different feeding processes for the timing of feeding.

[0111] Step 4: The mobile feeding device sends control commands to the target feeder according to the loading list and executes the interaction.

[0112] In this embodiment, after generating the loading list, the main control module of the mobile feeding device can autonomously plan a path and move to the location of the target loading machine. After completing precise parking, the mobile feeding device establishes a wireless communication connection with the target loading machine (e.g., establishing a handshake protocol via Wi-Fi, Bluetooth, or Industrial IoT protocols). After the connection is established, the mobile feeding device extracts key parameters from the loading list (including but not limited to feed type, target feeding weight, and proportioning requirements) and sends them to the target loading machine, and sends instructions to the target loading machine. These instructions serve as trigger signals, driving the target loading machine to automatically execute the corresponding physical loading actions according to the received loading parameters.

[0113] For example, the above instruction could be a "LOAD_START" instruction.

[0114] Optionally, in some embodiments, to ensure the fault tolerance and safety of the system, the mobile feeding device can also add a status confirmation mechanism before issuing the "LOAD_START" command to the target loading machine, thereby avoiding loading failures caused by loading machine occupancy or mismatch, and improving the reliability and execution efficiency of the feeding process.

[0115] Specifically, the mobile feeding device can first send a "LOAD_QUERY" command to the target feeder. This command is used to poll and verify the current working status of the target feeder (such as idle, in operation, or stopped due to malfunction) and the equipment matching degree (such as whether the hopper has the corresponding type of feed and whether the remaining amount meets the target weight).

[0116] For example, if the target feeder sends a confirmation message that it is "ready and has the capacity to match", the mobile feeder can send a "LOAD_START" command to the target feeder.

[0117] For example, if the target feeder reports "busy" or "mismatched", the moving feeder can trigger a waiting mechanism or reroute to another standby feeder.

[0118] The aforementioned closed-loop interaction mechanism can effectively avoid loading failures or even cross-contamination caused by the loading machine being occupied, material shortage in the hopper, or material mismatch, significantly improving the system reliability and multi-device collaborative execution efficiency of the fully unmanned feeding process.

[0119] Step 5: The target loading machine executes the loading task according to the instructions.

[0120] In this embodiment of the application, after receiving the loading parameters and instructions (e.g., LOAD_QUERY instructions) issued by the mobile feeding device, the target loading machine can perform the physical quantitative loading task of specific feed according to the instructions.

[0121] Specifically, during the loading process, the target loading machine continuously transmits dynamic loading (or feeding) progress data back to the mobile feeding device in real time based on the established soft bus near-field communication connection.

[0122] The loading progress data transmitted in real time by the target feeder may include, but is not limited to, the type of feed currently being loaded and the current loaded weight. Through this real-time status reporting mechanism, this system establishes a precise closed-loop monitoring network between the execution end (feeder) and the control end (mobile feeding equipment), ensuring that the mobile feeding equipment can have a global grasp of the dynamic process of physical operations and has the underlying data support for immediate intervention when abnormal material flow occurs. This ensures that the mobile feeding equipment, as a control node, can perform precise closed-loop monitoring of the physical operation process of the target feeder.

[0123] In this embodiment, the target loading machine integrates a high-precision internal metering unit (e.g., a weighing sensor or a dynamic flow meter). During the feeding process, when the internal metering unit detects and determines in real time that the "current loaded weight" reaches the "target weight" set in the instruction (or is within the preset allowable error range), the underlying controller of the target loading machine will immediately trigger the stop logic, automatically control the actuator to cut off the material flow, and stop the feeding action.

[0124] Specifically, after the target feeder stops moving, it actively sends a LOAD_STOP (feeding stopped / completed) status signal to the mobile feeding device via the aforementioned soft bus communication module. The successful return of this signal and the handshake signifies in the system control logic that the specific feed loading sub-task undertaken by the target feeder has been definitively completed in a closed loop. Simultaneously, this signal, as a crucial trigger mechanism, provides the final decision-making input for the mobile feeding device to update the verification status of the feeding list and advance the next stage of the system process (e.g., disconnecting or starting the internal mixing program).

[0125] Step 6: The mobile feeding device determines whether all feeding is complete based on the feeding list and the completed feeding tasks.

[0126] In this embodiment of the application, after receiving a single loading completion signal (such as the aforementioned LOAD_STOP signal) from the target loading machine, the main control system (or task scheduling module) of the mobile feeding device will synchronously update the local dynamic loading record and trigger a global task status verification algorithm.

[0127] Specifically, the mobile feeding device can compare all currently completed loading tasks (including the types of feed loaded and their cumulative weight) with the initially generated global loading list (e.g., overall formula data containing multiple feed ratios) to assess whether the overall formula has been met.

[0128] For example, if the mobile feeding device determines after verification that there are still unloaded items in the loading list (i.e., the types or weights of materials required for the current formula are not all available), the mobile feeding device can update the task pointer to the next unloaded item. At this point, the process jumps to and closes the loop to step 4, where the mobile feeding device replans its travel path to the next specific target loading machine, and repeats the status query, equipment networking, and loading command issuance process until all items in the list have been completely traversed.

[0129] For example, if the mobile feeding device verifies that all feed parameters loaded in the loading list have covered all the requirements in the loading list, that is, the global loading task is completed, the mobile feeding device will generate and store the internal "task completed" flag, then terminate the loading loop logic and trigger the state machine to transition to step 7.

[0130] Step 7: Move the mobile feeding equipment away from the feed warehouse.

[0131] In this embodiment of the application, after confirming that all loading tasks have been completed, the mobile feeding device can leave the feed warehouse.

[0132] For example, the underlying navigation control module of the mobile feeding equipment automatically plans the path to leave the current warehouse area based on the global operation map and the preset process flow, and drives the vehicle to the next operation node (such as the target shed).

[0133] Optionally, in some embodiments, in order to reduce the overall power consumption of the system, avoid cross-regional wireless signal interference, and prevent the occupation of public network resources, the communication module of the aforementioned mobile feeding device may also actively trigger the disconnection protocol of the soft bus module (or related near-field communication).

[0134] For example, a mobile feeding device can send an unbinding command (such as a DISCONNECT signal) to a target feeder that has previously established a connection, automatically deregistering and completely dismantling the network structure between the target feeder and the mobile feeding device.

[0135] The aforementioned automatic disconnection mechanism not only ensures the secure closure of the data transmission channel, but also ensures that the disconnected loading machine is quickly reset to a "completely idle" state, thus being ready to respond to the networking and loading requests of other mobile feeding equipment in the field at any time, realizing the industrial-grade scheduling requirements of seamless rotation and efficient collaboration among multiple vehicles and machines.

[0136] Step 220: The mobile feeding device acquires data from the environmental monitoring area within the target shed.

[0137] In this embodiment of the application, after the mobile feeding device obtains the feed to be fed from the feed warehouse, it can move from the feed warehouse to the target shed along the planned route.

[0138] For example, when a mobile feeding device, which serves as both a mobile operation node and the control decision-making brain, enters the near-field communication range (e.g., the coverage range of WiFi or Bluetooth) of a target shed, the mobile feeding device can trigger the sensing device detection mechanism of the target shed's environmental detection area.

[0139] For example, the mobile feeding device broadcasts a discovery command to the surrounding environment through its built-in soft bus communication module, automatically discovers the sensing devices within the environmental detection area of ​​the target shed, and actively sends a network request to the sensing devices within the environmental detection area.

[0140] For example, the sensing devices within the environmental monitoring area of ​​the aforementioned target shed may include, but are not limited to, surveillance cameras and environmental sensors.

[0141] For example, taking sensing devices such as surveillance cameras and environmental sensors as examples, such as... Figure 4 As shown, the aforementioned soft bus communication module is also deployed in the monitoring camera and environmental sensor. When the monitoring camera and environmental sensor located within the near-field range receive a broadcast from the mobile feeding device and a networking request, they will respond through their internally deployed soft bus communication module to establish the first near-field network.

[0142] Specifically, during the response process, the monitoring cameras and environmental sensors can proactively return their current status information to the mobile feeding device, which acts as the control node, based on the established short-range wireless communication connection (first near-field network). After obtaining the status information of the monitoring cameras and environmental sensors within the near-field range through the soft bus communication module, the mobile feeding device can verify the access requests of the monitoring cameras and environmental sensors. Once the monitoring cameras and environmental sensors pass the verification by the mobile feeding device, the mobile feeding device establishes a unified and interconnected network for feeding tasks with the monitoring cameras and environmental sensors in the target shed. This networking process breaks down the information silos previously operated independently by each sensing device, successfully connecting them into a real-time near-field linkage system.

[0143] For example, the aforementioned status information may include, but is not limited to: the device identifiers of the surveillance camera and the environmental sensor, and the operating status of the surveillance camera and the environmental sensor.

[0144] In this embodiment of the application, after the network between the mobile feeding device and the monitoring camera and environmental sensor is established, the devices in the network can start the aggregation and interaction of multi-source heterogeneous data.

[0145] For example, the aforementioned monitoring camera and environmental sensor can collect on-site perception data in real time and transmit it back to the mobile feeding device based on the unified data interaction capability of the soft bus communication module.

[0146] The following section, using Examples 1 and 2, provides a detailed description of the field perception data transmitted from the monitoring camera and environmental sensors to the mobile feeding device.

[0147] Example 1, such as Figure 4 As shown, the aforementioned monitoring camera is mainly responsible for collecting the current visual image data of the target shed in real time and sending the current visual image data to the mobile feeding device via a soft bus.

[0148] For example, the aforementioned current visual image data may include, but is not limited to: feed trough images, livestock feeding images, and shed passageway images. Among them, feed trough images are used to characterize the current state of remaining feed in the feed trough (i.e., the accumulation of uneaten feed after the previous feeding); livestock feeding images are used to characterize the current feeding activity of the livestock (i.e., the livestock's current appetite, health status, and the degree of feeding aggregation in the herd); shed passageway images are used to characterize the passageway's passability information (i.e., whether there are physical obstacles such as abnormal animal aggregation or equipment obstruction on the planned driving path of the feeding vehicle).

[0149] The aforementioned multi-dimensional visual image data provides intuitive visual data support for the subsequent dynamic reduction of planned feeding amounts, identification of abnormal behavior, and ensuring the safety of automated operation of mobile feeding equipment.

[0150] Example 2, such as Figure 4 As shown, the aforementioned environmental sensor is mainly responsible for collecting the current environmental data of the target shed in real time and sending the current environmental data to the mobile feeding device via a soft bus.

[0151] For example, the aforementioned current environmental data may include, but is not limited to, temperature and humidity data that characterize the microenvironment in which the livestock are located.

[0152] Optionally, in some embodiments, the aforementioned current environmental data may also include concentration data of harmful gases (such as ammonia, carbon dioxide, etc.) within the target shed.

[0153] The aforementioned current environmental data comprehensively reflects the climate comfort and air quality index within the sheds, and is used to assist mobile feeding equipment in assessing the stress response and changes in feeding behavior that livestock may experience under specific environmental conditions (such as high temperature, high humidity, or poor ventilation). This provides an objective environmental reference for the calculation of correction coefficients, dynamic adjustment of feeding amount and frequency in dynamic feeding strategies.

[0154] Step 230: The mobile feeding equipment determines the target feeding strategy for the target shed based on the data from the environmental monitoring area.

[0155] In this embodiment of the application, after the mobile feeding device obtains the on-site perception data of the environmental detection area collected in real time by the monitoring camera and environmental sensor in the target shed through the soft bus communication module, it can determine the target feeding strategy of the target shed based on the on-site perception data of the environmental detection area.

[0156] For example, the above-mentioned target feeding strategy may include at least one of the following: target single feeding amount, target feeding frequency, target discharge speed, and target execution action.

[0157] For example, a mobile feeding device can construct a multi-dimensional input dataset based on data from the environmental monitoring area and the pre-set feeding plan for the target shed.

[0158] For example, the multi-dimensional input dataset constructed above is shown in Table 1.

[0159] Table 1. Multidimensional Input Dataset

[0160] Referring to Table 1, the above input dataset may include: standard planned single feeding amount Q_plan_i, standard planned feeding frequency F_plan_i, standard planned feeding time T_plan_i, current feed trough remaining amount R_i, feeding activity A_i, passage passability flag L_i, temperature Temp_i, humidity Hum_i, and remaining amount of feed in the mobile feeding device M_i.

[0161] The aforementioned Q_plan_i, F_plan_i, and T_plan_i can also be called planned feeding task parameters. They are determined based on the preset feeding plan of the target shed numbered i. These parameters constitute the basic feeding benchmark for the target shed.

[0162] The aforementioned R_i, A_i, and L_i can also be referred to as the visual perception parameters (or on-site status data) of the target shed numbered i. These visual perception parameters are determined based on the current visual image data of the target shed collected by the monitoring camera. Specifically, R_i is determined based on the feed trough image collected by the monitoring camera and is used to represent the current amount of feed remaining in the feed trough; A_i is determined based on the livestock feeding image collected by the monitoring camera and is used to characterize the feeding desire and status of the livestock group in the target shed; and L_i is determined based on the shed passage image collected by the monitoring camera and is used to characterize whether there are physical obstacles such as abnormal animal aggregation or equipment obstruction on the travel path.

[0163] The aforementioned Temp_i and Hum_i can also be referred to as environmental perception parameters of the target shed numbered i. These environmental perception parameters are determined based on the current environmental data of the target shed collected by environmental sensors. Among them, Temp_i and Hum_i can be used to assess the microenvironmental thermal comfort of livestock within the target shed.

[0164] The aforementioned M_i is the current remaining material quantity of the mixing and feeding vehicle, which is fed back in real time by the vehicle's control system and is used to constrain the operational boundaries.

[0165] The input datasets shown in Table 1 above collectively form the data foundation for intelligent feeding strategy decision-making. Mobile feeding equipment can determine the target feeding strategy for the target shed based on these multi-source input data.

[0166] There are multiple ways to determine the target feeding strategy for the target shed based on the input dataset. This application does not limit the specific implementation of this method. The following describes three possible determination methods in detail with reference to implementation methods 1 to 3.

[0167] Implementation method 1: The mobile feeding device determines the above target feeding strategy according to the preset feeding strategy formula.

[0168] For example, in implementation method 1, the mobile feeding device can calculate based on the above input data using a preset feeding strategy formula to generate feeding strategy 1, and directly use feeding strategy 1 as the target feeding strategy for the target shed. This implementation method has the characteristics of low computational energy consumption, high determinism, and fast response speed.

[0169] The following describes in detail the specific implementation process of determining the target feeding strategy based on the preset feeding strategy formula, in conjunction with steps a-e.

[0170] Step a: Calculate the initial feeding requirement for a single feeding at the target shed.

[0171] For example, the mobile feeding device first obtains the standard planned single feeding amount Q_plan_i from the input dataset, and subtracts the current remaining amount of feed in the target shed R_i from Q_plan_i to calculate the initial single feeding requirement Q_need_i for the target shed.

[0172] For example, the mobile feeding device can calculate the initial demand Q_need_i for a single feeding according to the following formula (1).

[0173] Q_need_i = max(0, Q_plan_i - R_i)(1) The above formula (1) ensures at the underlying logic that when there is too much residual feed from the previous round, the initial demand for a single feeding can be automatically reduced, thus preventing serious feed accumulation and spoilage in the feed trough from the source.

[0174] Step b: Adjust the initial feeding requirement for a single feeding based on the livestock's feeding behavior and the environmental conditions of the target shed, and calculate the feeding amount for a single feeding.

[0175] For example, in order to cope with the complex physiological feeding state of livestock and the dynamic fluctuations of the microenvironment within the target shed, this application embodiment introduces a feeding amount correction coefficient K_i. The mobile feeding device can correct the Q_need_i calculated above according to K_i to obtain the single feeding amount Q_rule_i determined based on the rule formula.

[0176] For example, the mobile feeding device can calculate the above single feeding amount Q_rule_i according to the following formula (2).

[0177] Q_rule_i = Q_need_i × K_i (2) For example, the mobile feeding device described above can determine the correction coefficient K_i based on the actual on-site feeding activity and environmental conditions of the livestock in the target shed, in conjunction with Table 2.

[0178] Table 2. Values ​​of the correction coefficient K_i

[0179] Referring to Table 2, for example, if the livestock in the target shed are actively eating, have little leftover feed, and the environment is normal, then these livestock have a strong appetite. The correction coefficient K_i ranges from 1.0 to 1.2, and the corresponding strategy is to maintain the initial demand for feed or supplement with appropriate feed. Alternatively, if the livestock in the target shed are eating normally and have a normal amount of leftover feed, then these livestock are in a standard ideal state, and the correction coefficient K_i ranges from 1.0. The corresponding strategy is to feed according to the initial demand. Furthermore, if the livestock in the target shed show decreased eating activity, excessive leftover feed, or high temperature and humidity, it indicates that the livestock are experiencing decreased appetite due to environmental stress or reduced digestive function. The correction coefficient K_i ranges from 0.5 to 0.9, and the corresponding strategy is to reduce the amount of feed based on the initial demand to achieve precise feed control and eliminate waste. For example, if the on-site situation is that there are blocked passages, abnormal animal gatherings, or equipment malfunctions in the target shed, it is considered a serious on-site abnormality. The correction coefficient K_i is set to 0, and the corresponding strategy means to suspend feeding, thereby raising the safety control threshold of the system.

[0180] Step c: Adjust the standard feeding frequency according to the livestock's feeding behavior and the environmental conditions of the target shed, and determine the frequency of each feeding.

[0181] For example, in order to cope with the complex physiological feeding state of livestock and the dynamic fluctuations of the microenvironment within the target shed, this application embodiment also introduces a frequency adjustment amount F_adjust_i. The mobile feeding device can modify the above-mentioned standard planned feeding frequency F_plan_i according to F_adjust_i to obtain the single feeding frequency F_rule_i determined based on the rule formula.

[0182] For example, the mobile feeding device can calculate the above feeding frequency F_rule_i according to the following formula (3).

[0183] F_ rule _i = F_plan_i + F_adjust_i (3) For example, the mobile feeding device described above can determine the frequency adjustment amount F_adjust_i based on the actual on-site feeding activity and environmental conditions of the livestock in the target shed, in conjunction with Table 3.

[0184] Table 3. Values ​​of frequency adjustment amount F_adjust_i

[0185] Referring to Table 3, for example, if the livestock in the target shed are actively eating, have little uneaten feed, and the environment is normal, then these livestock have a strong appetite, and the frequency adjustment value F_adjust_i is 0 or +1. The corresponding strategy is to maintain the planned feeding frequency or increase the number of feedings. Alternatively, if the livestock in the target shed are eating normally and have normal uneaten feed, then these livestock are in a standard ideal state, and the frequency adjustment value F_adjust_i is 0. The corresponding strategy is to feed according to the planned feeding frequency. Furthermore, if the livestock in the target shed show decreased eating activity or excessive uneaten feed, it indicates that the livestock's appetite is reduced due to decreased digestive function. The frequency adjustment value F_adjust_i is 0 or -1, and the corresponding strategy is to reduce the number of feedings or maintain the same frequency based on the planned feeding frequency. For example, if the environment inside the target shed is hot and humid, indicating that livestock are experiencing decreased appetite due to environmental stress, the frequency adjustment value F_adjust_i is +1. This strategy means increasing the feeding frequency. This rule aims to prevent feed from fermenting and spoiling due to prolonged exposure to high temperature and humidity through a "small amounts, multiple times" rhythm, while simultaneously stimulating livestock to eat. Alternatively, if the environment inside the target shed experiences blocked passageways, abnormal animal aggregation, or equipment malfunction, the frequency adjustment value F_adjust_i is +1, meaning supplementing the feeding frequency for compensatory feeding after the abnormality is resolved.

[0186] It should be understood that the aforementioned correction coefficient K_i and frequency adjustment amount F_adjust_i can correspond to the correction coefficients mentioned above.

[0187] Step d: Determine the single discharge rate D_rule_i based on the single feeding amount Q_rule_i of the target shed and the standard planned feeding time T_plan_i.

[0188] For example, the mobile feeding device can calculate the above single discharge speed D_rule_i according to the following formula (4).

[0189] D_rule_i = Q_ rule_i / T_plan_i (4) Step e: Generate a strategy suggestion action Action_rule_i based on the single feeding amount Q_rule_i, the single feeding frequency F_rule_i, and the on-site status of the target shed.

[0190] In this embodiment of the application, the mobile feeding device can determine the Action_rule_i by searching the value table of Action_rule_i shown in Table 4 based on the single feeding amount Q_rule_i, the single feeding frequency F_rule_i and the on-site status of the target shed, so as to match the corresponding control behavior instruction.

[0191] It should be understood that in implementation method 1, Q_rule_i corresponds to the target single feeding amount mentioned above, F_rule_i corresponds to the target feeding frequency mentioned above, D_rule_i corresponds to the target discharge speed mentioned above, and Action_rule_i corresponds to the target execution action mentioned above.

[0192] Table 4 Values ​​of Action_i

[0193] Referring to Table 4, if the single feeding amount Q_rule_i is close to (or equal to) the standard planned single feeding amount Q_plan_i, and the site conditions are normal, then the control behavior command issued by the mobile feeding device is normal feeding; if the single feeding amount Q_rule_i is greater than the standard planned single feeding amount Q_plan_i, and the site conditions are normal, then the control behavior command issued by the mobile feeding device is incremental feeding; if the single feeding amount Q_rule_i is less than the standard planned single feeding amount Q_plan_i, and the single feeding frequency F_rule_i is the same as the standard planned feeding frequency F_plan_i, then the control behavior command issued by the mobile feeding device is... The action command is to reduce feeding; if the single feeding amount Q_rule_i is less than the standard planned single feeding amount Q_plan_i, and the single feeding frequency F_rule_i is greater than the standard planned feeding frequency F_plan_i, then the control action command issued by the mobile feeding device is to feed in small amounts and multiple times; if the passage is blocked, animals gather abnormally, or the equipment malfunctions, then the control action command issued by the mobile feeding device is to suspend feeding; if the calculated single feeding amount Q_rule_i is greater than the current remaining feed amount M_i of the mobile feeding device (i.e., a physical boundary conflict occurs due to insufficient remaining feed), then the system automatically generates an instruction to return to the feed warehouse for replenishment.

[0194] It should be understood that the single feeding amount Q_rule_i, single feeding frequency F_rule_i, single discharge speed D_rule_i, and the specific execution action Action_rule_i together constitute the formula strategy parameter set P_rule. That is, P_rule is the core parameter set of the above feeding strategy 1, P_rule={Q_rule_i, F_rule_i, D_rule_i, Action_rule_i}. Among them, the execution action Action_rule_i is a highly abstract and semantic representation of the cascaded calculation results of Q_rule_i and F_rule_i. The mobile feeding device automatically matches macro-level behavioral instructions such as "normal feeding", "reduced feeding", or "small amount, multiple feedings" by logically verifying Q_rule_i and F_rule_i. In other words, the Action_rule_i gives the feeding strategy 1 a specific behavioral soul, allowing the mobile feeding device to clearly perceive the underlying management intent of the current strategy (such as dealing with the risk of high-temperature fermentation by "small amounts and multiple times").

[0195] Implementation Method 2: The mobile feeding device determines the above-mentioned target feeding strategy by calling the AI ​​large model.

[0196] For example, in implementation method 2, the mobile feeding device can generate feeding strategy 2 by calling the AI ​​large model based on the above input data, and directly use feeding strategy 2 as the target feeding strategy for the target shed. This implementation method has the characteristics of low computational energy consumption, strong determinism, and fast response speed.

[0197] It should be understood that in implementation method 2, in order to overcome the limitations of traditional fixed formulas in handling complex multivariate communities, long-term behavioral characteristics, and implicit breeding patterns, the mobile feeding device can determine the above-mentioned target feeding strategy by calling an AI large model.

[0198] In practical applications, this AI big model typically employs a multimodal neural network or a time series prediction big model with high generalization ability, which can deeply understand the cross-modal correlation between images and environmental values, and achieve accurate perception and intelligent decision-making for complex livestock farming scenarios.

[0199] This application does not specifically limit the type of the above-mentioned large AI model in its embodiments. Several common types are listed below.

[0200] 1) Multimodal fusion large model: With visual Transformer and temporal encoder as the core architecture, it can simultaneously process multi-source heterogeneous data such as images, videos, temperature, humidity, and gas concentration. It can mine the deep correlation between visual features and environmental parameters through cross-modal attention mechanism, and is suitable for livestock behavior recognition and comprehensive environmental status assessment.

[0201] 2) Large-scale time series forecasting models: Represented by temporal fusion transformer (TFT) and Informer, these models specialize in long-term modeling and trend forecasting of continuously changing time series data in the breeding environment (such as temperature and humidity curves, feed consumption trends, and changes in livestock weight) to support forward-looking management decisions.

[0202] 3) Visual Language Large Model: Based on technologies such as CLIP and LLaVA, it has dual capabilities of image understanding and natural language reasoning. It can transform images captured by cameras into structured semantic descriptions and combine them with user commands to complete tasks such as anomaly alarms and aquaculture plan generation.

[0203] 4) Domain-specific pre-trained large model: Based on the general large model, the model is fine-tuned using massive labeled data in the field of animal husbandry, so that the model has higher accuracy and reliability in specific tasks such as determining livestock feeding strategies.

[0204] To adapt to different computing resources and real-time requirements, this application does not impose specific limitations on the deployment method of the above-mentioned large AI model. It can be flexibly selected according to the actual application scenario. Several possible implementation methods are listed below.

[0205] For example, it can be deployed locally in the high-performance edge computing unit of a mobile feeding device, and the entire model inference process is completed at the vehicle end without relying on an external network connection, thereby achieving microsecond-level local response in a network-off environment and ensuring the continuity and real-time nature of the feeding operation.

[0206] For example, it can be distributed and deployed in local edge servers in various livestock sheds. The edge nodes of each livestock shed can work together to complete the data aggregation and model inference within the region, so as to realize centralized detection and real-time control of livestock status in a specific region, taking into account both response speed and regional management efficiency.

[0207] For example, it can be deployed in the cloud, leveraging the massive computing resources of the cloud to perform long-term historical data analysis and macro-strategic reasoning across different livestock sheds. This could involve making global optimization decisions regarding feed consumption trends, livestock growth curves, and disease transmission risks across the entire ranch, providing management with intelligent decision support from a holistic perspective.

[0208] Optionally, in some embodiments, the above three deployment methods can also be combined to form a "device-edge-cloud" collaborative architecture, where the vehicle (e.g., a mobile feeding device) is responsible for immediate response, the edge is responsible for regional control, and the cloud is responsible for global optimization, thereby achieving the best balance between real-time performance, reliability, and decision-making depth.

[0209] The following is a detailed description of the specific implementation process of the mobile feeding device determining the above-mentioned target feeding strategy by calling the AI ​​large model.

[0210] For example, the information input by the aforementioned mobile feeding device to the AI ​​big model may include, but is not limited to: status input information, environmental input information, device input information, task input information, experience input information, etc.

[0211] 1) Status Input Information: This includes the current remaining feed level in the trough R_i, livestock feeding activity level A_i, and image results of abnormal behaviors such as passage obstruction and animal aggregation. Its purpose is to provide the AI ​​model with instantaneous visual status features of the current feeding operation site, enabling it to accurately assess the livestock's real-time feeding desire and safety boundaries.

[0212] 2) Environmental input information: This includes the current temperature (Temp_i) and humidity (Hum_i) of the target shed. Its purpose is to provide the AI ​​model with the climate stress factors of the livestock's microenvironment, so that the AI ​​model can quantify the implicit impact of the environment on the livestock's digestive function.

[0213] 3) Equipment input information: This refers to the current remaining material quantity M_i of the mobile feeding device itself. Its function is to provide the physical capability boundary of the current control subject to the AI ​​model, preventing the AI ​​model from issuing invalid instructions that exceed the remaining hardware load.

[0214] 4) Task Input Information: This includes the target shed number i, the standard planned single feeding amount Q_plan_i, the standard planned feeding frequency F_plan_i, and the standard planned feeding time T_plan_i. Its purpose is to provide the AI ​​model with baseline operational guidelines for this task, ensuring that it has a clear framework for inference.

[0215] 5) Experience input information: This includes feeding records from multiple rounds of the target shed, historical anomaly records, manual adjustment records, and breeding management rules for specific breeds. Its purpose is to provide the AI ​​model with long-term temporal memory and expert assets, enabling it to overcome the limitations of single-perception and perform forward-looking reasoning by combining historical patterns.

[0216] For example, the aforementioned large AI model directly outputs a structured feeding strategy 2 by performing deep attention mechanism fusion and nonlinear feature mapping on the above input information.

[0217] Specifically, the feeding strategy 2 output by the aforementioned AI model includes, but is not limited to, the following information: a summary of the target shed's status, parameter suggestions, and explanations of the reasons for strategy adjustments.

[0218] The aforementioned target shed status summary is used to provide a high-level overview of the current deep-seated complex health and environmental status of the target shed, serving to provide semantic status diagnosis for the management of the entire feeding system. For example, "The current temperature and humidity index of the target shed is high, livestock are experiencing mild heat stress, their feeding activity is fluctuating and decreasing, and there is excessive uneaten feed in the troughs."

[0219] The above-mentioned strategy suggestions (Action_ai) include macro-level decision-making actions such as normal feeding, reduced feeding, increased feeding, delayed feeding, small-volume multiple feeding, pausing feeding, or returning to the warehouse for replenishment. Their role is to serve as the highest execution semantic of the mobile feeding equipment control state machine, directly locking in the next core operation direction of the mobile feeding equipment.

[0220] The above parameter suggestions directly provide the AI ​​strategy parameter set P_ai obtained through AI large-scale model inference. Where P_ai = {Q_ai_i, F_ai_i, D_ai_i, Action_ai_i}, Q_ai_i represents the AI-recommended single feeding amount, F_ai_i represents the AI-recommended feeding frequency, D_ai_i represents the AI-recommended single discharge speed, and Action_ai_i represents the AI-recommended action for this strategy.

[0221] It should be understood that in implementation method 2, Q_ai_i corresponds to the target single feeding amount mentioned above, F_ai_i corresponds to the target feeding frequency mentioned above, D_ai_i corresponds to the target discharge speed mentioned above, and Action_ai_i corresponds to the target execution action mentioned above.

[0222] For example, the strategy suggestions (Action_ai) recommended by the AI ​​include, but are not limited to: normal feeding, reduced feeding, increased feeding, delayed feeding, small-volume frequent feeding, pausing feeding, or returning to the warehouse for replenishment, etc. Their role is to serve as the highest execution semantic of the mobile feeding equipment's control state machine, directly determining the next core operational direction of the mobile feeding equipment.

[0223] The explanation of the reasons for the above strategy adjustments provides a technical attribution explanation, which serves to clarify the underlying technical logic of the strategy modification to the system or human reviewers, thereby achieving traceability and transparency in AI decision-making. For example, "Excessive leftover feed is mainly related to decreased feed intake under high temperature conditions. It is not advisable to continue feeding according to the standard amount. It is recommended to reduce the amount per feeding and increase the frequency."

[0224] In implementation method 2 described above, after receiving the strategy parameters output by the AI ​​big data model, the mobile feeding device directly issues and executes feeding strategy 2, which includes the above four output items, as the target feeding strategy for the target shed. This intelligent strategy generation mechanism, driven by AI big data model data, can accurately capture subtle changes in livestock behavior caused by environmental stress that are difficult to cover with traditional fixed formulas. This significantly reduces the cost of manual intervention while greatly improving the adaptability of strategies and the efficiency of precise feeding in abnormal scenarios.

[0225] Implementation method 3: The mobile feeding device determines the above-mentioned target feeding strategy through preset rule formulas and AI large model.

[0226] For example, in order to balance the "high security and high certainty baseline" of traditional rule formulas with the "high flexibility and high foresight ceiling" of AI big data models, in implementation method 3, the mobile feeding device adopts a dual-track parallel intelligent decision-making logic. By deeply fusing the control scheme calculated based on the preset rule formula (e.g., the aforementioned feeding strategy 1 or the first feeding strategy) with the intelligent suggestions inferred based on the AI ​​big data model (e.g., the aforementioned feeding strategy 2 or the second feeding strategy), the target feeding strategy for the target shed is determined.

[0227] The aforementioned dual-track verification mechanism can effectively prevent large AI models from "illusioning" or issuing out-of-bounds instructions in extreme long-tail scenarios, ensuring the robustness of the entire feeding linkage system control.

[0228] One possible implementation is, such as Figure 5 As shown, the mobile feeding device can obtain the aforementioned parameter set P_rule and parameter set P_ai, calculate the deviation between the parameters in P_rule and the parameters in P_ai, and determine the target feeding strategy based on the deviation between the parameters. Specifically, parameter set P_rule is the set of formula strategy parameters determined by the mobile feeding device according to the preset rule formula in implementation method 1, P_rule = {Q_rule_i, F_rule_i, D_rule_i, Action_rule_i}. Parameter set P_ai is the set of AI strategy parameters obtained by the mobile feeding device through inference using a large AI model in implementation method 2, P_ai = {Q_ai_i, F_ai_i, D_ai_i, Action_ai_i}.

[0229] For example, the following describes in detail the specific implementation process of calculating the deviation between the parameters in the formula policy parameter set P_rule and the parameters in the AI ​​policy parameter set P_ai.

[0230] For example, the deviation between the parameters in the formula strategy parameter set P_rule and the parameters in the AI ​​strategy parameter set P_ai can be calculated according to formulas (5)-(7).

[0231] ΔQ_i=|Q_ai_i - Q_rule_i| / max(Q_rule_i, ε)(5) ΔF_i=|F_ai_i - F_rule_i | / max(F_rule_i, ε) (6) ΔD_i=|D_ai_i - D_rule_i| / max(D_rule_i, ε) (7) Where ΔQ_i represents the deviation of the amount fed per feeding, ΔF_i represents the deviation of the feeding frequency, ΔD_i represents the deviation of the feeding speed per feeding, and ε is a preset minimum value to prevent the denominator from being zero.

[0232] The following section describes in detail the specific implementation method of determining the target feeding strategy based on the deviation between the above parameters and the fusion mechanism.

[0233] Example 1: such as Figure 5 As shown, when the calculated deviation of a single feeding amount ΔQ_i is less than the preset deviation threshold of the feeding amount, the deviation of the feeding frequency ΔF_i is less than the preset deviation threshold of the feeding frequency, the deviation of the single feeding speed ΔD_i is less than the preset deviation threshold of the feeding speed, and the strategy suggestion action Action_ai_i given by the AI ​​big model is consistent with the execution action Action_rule_i given by the rule formula, the mobile feeding device can directly adopt the AI ​​strategy result output by the AI ​​big model as the target feeding strategy of the target shed and issue it for execution, so that the feeding operation can seamlessly enjoy the fine-tuning benefits brought by the big model.

[0234] Example 2: such as Figure 5 As shown, when any one or more of the calculated ΔQ_i, ΔF_i, and ΔD_i exceed the corresponding preset deviation threshold, and the strategy suggestion action Action_ai_i given by the AI ​​big model is consistent with the execution action Action_rule_i given by the rule formula, it can be understood that the AI ​​big model has made a significant forward-looking and flexible adjustment based on rich historical aquaculture experience assets, but has not exceeded the safety business red line. In this case, a weighted fusion mechanism can be introduced to weight and fuse the parameters in the formula strategy parameter set P_rule and the AI ​​strategy parameter set P_ai to obtain a comprehensive feeding strategy (i.e., feeding strategy 3), and this fused feeding strategy 3 is delivered as the final target feeding strategy to the mobile feeding equipment for execution.

[0235] For example, the parameters in P_rule and P_ai can be weighted and fused according to formula (8) to obtain the final parameter set P_final after fusion.

[0236] P_final=λ × P_ai + (1 - λ) × P_rule (8) Where λ is the AI ​​policy weight, and its value ranges from 0 to 1.

[0237] The AI ​​strategy weights mentioned above can be pre-set empirical values ​​or manually confirmed values. This application does not specifically limit them.

[0238] Example 3: such as Figure 5 As shown, when any one or more of the calculated ΔQ_i, ΔF_i, and ΔD_i exceed the corresponding preset deviation threshold, and the strategy suggestion action Action_ai_i given by the AI ​​big model is inconsistent with the execution action Action_rule_i given by the rule formula (for example, the rule formula requires "pause feeding" based on the safety boundary calculation of the hopper, while the AI ​​big model gives "incremental feeding" due to long-term prediction), it can be determined that there may be an extreme long-tail anomaly at the current work site or that the big model's inference has generated a control illusion. In this case, to ensure safety, the conflict determination process can be initiated.

[0239] For example, in a typical conflict scenario, the rule formula, based on real-time calculations of the feed trough's physical safety boundaries, determines that the current feed level is nearing its limit and explicitly requires "pausing feeding." However, the AI ​​model, based on long-term predictions of livestock's recent feeding trends and growth cycles, determines that the current feeding window is still valid and suggests "incremental feeding." The two instructions are completely opposite. In this situation, the system cannot automatically determine which is more reliable. The current work site may indeed have extreme long-tail anomalies beyond the historical data coverage (such as sudden group stress in livestock causing data distortion in the feed trough), or the AI ​​model may have experienced a control illusion during reasoning, making a seemingly reasonable but actually dangerous decision based on incorrect contextual associations.

[0240] Specifically, to ensure operational safety, the following tiered response mechanisms can be triggered.

[0241] 1) Local level: The mobile feeding device automatically enters a safe standby state, suspends the execution of feeding instructions from either party, and at the same time, a conflict alarm pops up on the vehicle-mounted human-machine interface to remind the operator to pay attention.

[0242] 2) Cloud level: The cloud management platform receives the security alert synchronously and uploads the complete context data of this conflict, including but not limited to: the specific values ​​of the current ΔQ_i, ΔF_i, and ΔD_i and their deviation from the threshold, the complete inference chain of the AI ​​large model output Action_ai_i, the calculation basis of the rule engine output Action_rule_i, and environmental values ​​and image snapshots before and after the conflict occurred.

[0243] 3) Manual Arbitration: Online monitoring experts will conduct a comprehensive assessment and ruling based on the complete data mentioned above. Experts can choose to adopt AI strategies, rule-based strategies, or manually adjust and generate new feeding plans.

[0244] In this embodiment of the application, the feeding strategy that has been finally confirmed by a person or adjusted manually can be marked as the "target feeding strategy" and officially issued to the mobile feeding device for execution through a secure channel.

[0245] Optionally, in some embodiments, the aforementioned conflict events and adjudication results may be recorded and archived as labeled samples for subsequent model iterations and rule optimizations.

[0246] Step 240: The mobile feeding equipment performs the feeding task on the livestock in the target shed according to the target feeding strategy.

[0247] In this embodiment of the application, after the mobile feeding device finally determines the target feeding strategy (i.e., the aforementioned feeding strategy 1, feeding strategy 2 or the weighted fusion feeding strategy 3), it unpacks the strategy and converts it into specific low-level execution instructions, thereby starting the feeding task in the target shed.

[0248] For example, the mobile feeding device first controls its walking mechanism to switch to the corresponding operating state (e.g., "normal feeding", "incremental feeding", or "reduced feeding") according to the execution action (Action_i) in the target feeding strategy. Subsequently, the vehicle body control system of the mobile feeding device dynamically adjusts the operating state of the underlying drive hardware according to the quantitative control parameters given in the strategy.

[0249] Specifically, during the feeding process, on the one hand, the mobile feeding device adjusts the opening of its discharge gate and the rotation speed of the stirring auger motor in real time according to the discharge speed determined by the target feeding strategy, to ensure that the weight of feed sprayed per unit time precisely matches the needs of the target shed. On the other hand, the mobile feeding device controls its round trips or waiting intervals at the target shed according to the final feeding frequency in the target feeding strategy. For example, when the action is defined as "feeding in small amounts multiple times", the mobile feeding device controls itself to pass through the target shed at a faster speed and a smaller gate opening to reduce the amount of feed per feeding, and returns to the target shed again after a preset interval to supplement feeding, thereby completing the total feeding amount of the zone in batches and at high frequency. If, during the feeding operation, the system's visual perception input detects physical safety red lines such as passage obstruction, or if the calculated final feeding amount is greater than the current remaining material amount M_i on the vehicle, the mobile feeding device will interrupt the current feeding logic and instead execute a safety boundary crossing action of "pausing feeding" or "returning to the warehouse for replenishment".

[0250] Optionally, in some embodiments, once the mobile feeding device successfully and accurately delivers the final feeding amount to the feed trough in the target shed, the feeding task in the current target shed is considered complete. At this point, the mobile feeding device packages all key business characteristics involved in this round of operation, including initial multi-source sensing input data, calculation deviations from intermediate strategy fusion, final issued strategy control parameters, and the device status at the end of feeding, into a new set of experiential feature data and re-inputs it into the AI ​​model for online calibration and adaptive incremental training. This provides data asset support for strategy optimization in the next round of operation. This self-correcting closed-loop feedback mechanism enables the model's strategy recommendation accuracy to continuously increase with the accumulation of operation time.

[0251] Optionally, after completing the feeding task for the current target shed or zone, the mobile feeding device will also consider the global feeding plan and the current remaining feed amount to determine whether it needs to enter the next target shed or zone to continue feeding. If so, it will move to the next area and restart the near-field discovery and networking process. If the entire plan is completed or there is insufficient remaining feed, it will leave the shed area.

[0252] Alternatively, in some embodiments, such as Figure 5As shown, if the mobile feeding device completes its feeding task in the target shed and then leaves the target shed, it automatically disconnects from the network of cameras and environmental sensors within the target shed. For example, as the mobile feeding device moves physically, when its built-in soft bus communication module detects that the wireless signal strength between it and the monitoring cameras and environmental sensors in the target shed has attenuated to below a preset disconnection threshold, or when it receives a shed task termination command from the vehicle control system, it automatically triggers the soft bus de-networking mechanism.

[0253] Specifically, the mobile feeding device sends a network handshake signal to the cameras and environmental sensors in the target shed via a soft bus communication network, securely releasing the network communication link and buffer channel, thereby automatically terminating the task networking relationship with the monitoring cameras and environmental sensors in the target shed.

[0254] The aforementioned de-networking process not only frees up the wireless communication bandwidth and computing resources of the sensing device itself, enabling it to re-enter a low-power standby state or wait to serve the next operating vehicle, but also clears the soft bus peer node list of the mobile feeding device itself, preventing invalid topologies from residing in memory. This ensures that the mobile feeding device has efficient and clean near-field dynamic re-networking capabilities when it enters the next unknown shed.

[0255] In the above technical solution, the automatic discovery and dynamic networking capabilities of the soft bus communication module can connect mobile feeding devices, feeders, cameras, and environmental sensors in large-scale feed feeding scenarios into a near-field linkage system. The mobile feeding devices can automatically connect to the feeders, cameras, and environmental sensors in the current area during mobile operations, achieving real-time near-field linkage and reducing the difficulty of integrating equipment from different manufacturers. Simultaneously, based on the unified data interaction capabilities of the soft bus, the mobile feeding devices can aggregate on-site sensing data acquired by different devices, dynamically adjust and execute feeding strategies, improve the automation and real-time coordination capabilities of feed loading and dispensing processes, reduce manual intervention costs, and improve feed dispensing efficiency, strategy adaptability, and control security in abnormal scenarios.

[0256] The above text combined Figures 1 to 5 The method provided in the embodiments of this application is described in detail below. Figures 6-7 The embodiments of the apparatus of this application are described in detail below. It should be understood that the descriptions of the method embodiments correspond to the descriptions of the apparatus embodiments; therefore, any parts not described in detail can be referred to the foregoing method embodiments.

[0257] Figure 6This is a schematic block diagram of an intelligent aquaculture device 600 based on a soft bus, provided in an embodiment of this application. The device 600 can be implemented through software, hardware, or a combination of both. The device 600 provided in this embodiment is deployed in a mobile feeding device, which also includes a soft bus communication module. This device 600 can implement the functions described in the embodiment of this application. Figure 2 The steps performed by the mobile feeding device in the illustrated method flow include: a creation module 610, an acquisition module 620, a determination module 630, and an execution module 640. The creation module 610 automatically establishes a second near-field network with the target feeder via a soft bus communication module within the near-field range of the feed warehouse before the mobile feeding device enters the target shed. The target feeder is equipped with a soft bus communication module. The acquisition module 620 sends a feeding control command to the target feeder via the second near-field network, causing the target feeder to feed the mobile feeding device according to the feeding list. Feed; the creation module 610 is also used to automatically establish a first near-field network with the sensing device deployed in the target shed through the soft bus communication module within the near-field range of the target shed after the mobile feeding device completes the feed dispensing; the sensing device is equipped with a soft bus communication module; the acquisition module 620 is also used to acquire the field sensing data collected by the sensing device through the first near-field network; the determination module 630 is used to determine the target feeding strategy for the target shed based on the field sensing data and the preset feeding plan; the execution module 640 is used to execute the feed dispensing task according to the target feeding strategy.

[0258] Optionally, the target feeding strategy includes at least one of the following: target single feeding amount, target feeding frequency, target discharge speed, and target execution action.

[0259] Optionally, the determining module 630 is specifically used to: determine an input dataset based on the field sensing data and the feeding plan, wherein the input dataset includes the planned single feeding amount, the planned feeding frequency, the field status data of the target shed, and the remaining amount of material in the mobile feeding device; and determine the target feeding strategy based on the input dataset.

[0260] Optionally, the determining module 630 is specifically used to: calculate the target feeding strategy based on the input dataset and the preset feeding strategy formula.

[0261] Optionally, the determining module 630 is specifically used to: use a correction coefficient determined based on the on-site status data of the target shed to correct the planned single feeding amount and / or the planned feeding frequency to obtain the target single feeding amount and the target feeding frequency; and calculate the target discharge rate based on the target single feeding amount and the planned feeding time in the feeding plan.

[0262] Optionally, the determining module 630 is specifically used to: call an artificial intelligence (AI) big model based on the input dataset to obtain the target feeding strategy, wherein the input information of the AI ​​big model includes the input dataset, and the output information of the AI ​​big model includes the target feeding strategy.

[0263] Optionally, the determining module 630 is specifically used to: calculate a first feeding strategy based on the input dataset and a preset feeding strategy formula; call an AI large model based on the input dataset to obtain a second feeding strategy; and perform strategy fusion on the first feeding strategy and the second feeding strategy to obtain the target feeding strategy.

[0264] Optionally, the determining module 630 is specifically used to: calculate the deviation value between the first feeding strategy and the second feeding strategy in terms of control parameters and / or execution actions; and determine the target feeding strategy based on the deviation value using the corresponding fusion rule.

[0265] Optionally, the determining module 630 is specifically used to: determine the second feeding strategy as the target feeding strategy when the deviation value is less than a preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent; or when the deviation value is greater than or equal to the preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent, perform weighted fusion of the control parameters in the first feeding strategy and the second feeding strategy to obtain the target feeding strategy; or when the execution actions of the first feeding strategy and the second feeding strategy are inconsistent, trigger an alarm prompt to obtain the target feeding strategy for manual confirmation.

[0266] Optionally, the sensing device includes a surveillance camera and an environmental sensor, and the on-site sensing data includes at least one of the following: temperature or humidity collected by the environmental sensor, images of feed troughs, images of livestock feeding, and images of shed passages collected by the surveillance camera.

[0267] The device 600 here can be embodied in the form of a functional module. The term "module" here can be implemented in software and / or hardware, without specific limitations.

[0268] For example, a "module" can be a software program, a hardware circuit, or a combination of both that implements the above functions. For instance, the following description uses module 610 as an example to illustrate its implementation. Similarly, the implementation of other modules, such as module 620, module 630, and module 640, can refer to the implementation of module 610.

[0269] Creation module 610, as an example of a software functional unit, may include code running on a computing instance. The computing instance may include at least one of a physical host (computing device), a virtual machine, or a container. Further, the aforementioned computing instance may be one or more. For example, creation module 610 may include code running on multiple hosts / virtual machines / containers. It should be noted that the multiple hosts / virtual machines / containers used to run the code may be distributed within the same region or in different regions. Further, the multiple hosts / virtual machines / containers used to run the code may be distributed within the same availability zone (AZ) or in different AZs, each AZ including one or more geographically proximate data centers. Typically, a region may include multiple AZs.

[0270] Similarly, multiple hosts / virtual machines / containers used to run this code can be distributed within the same Virtual Private Cloud (VPC) or across multiple VPCs. Typically, a VPC is set up within a region. Communication between two VPCs within the same region, as well as between VPCs in different regions, requires a communication gateway to be set up within each VPC to enable interconnection between VPCs.

[0271] Creation module 610, as an example of a hardware functional unit, may include at least one computing device, such as a server. Alternatively, creation module 610 may also be a device implemented using an application-specific integrated circuit (ASIC) or a programmable logic device (PLD). The aforementioned PLD may be implemented using a complex programmable logical device (CPLD), a field-programmable gate array (FPGA), generic array logic (GAL), or any combination thereof.

[0272] The multiple computing devices included in creation module 610 can be distributed in the same region or in different regions. Similarly, the multiple computing devices included in creation module 610 can be distributed in the same Availability Zone (AZ) or in different AZs. Likewise, the multiple computing devices included in creation module 610 can be distributed in the same Virtual Private Cloud (VPC) or in multiple VPCs. These multiple computing devices can be any combination of computing devices such as servers, ASICs, PLDs, CPLDs, FPGAs, and GALs.

[0273] Therefore, the modules of the various examples described in the embodiments of this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0274] It should be noted that the above embodiments of the device, when executing the above methods, are only illustrative examples of the division of functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. For example, the creation module 610 can be used to execute any step in the above methods, the acquisition module 620 can be used to execute any step in the above methods, the determination module 630 can be used to execute any step in the above methods, and the execution module 640 can be used to execute any step in the above methods. The steps implemented by the creation module 610, acquisition module 620, determination module 630, and execution module 640 can be specified as needed. By implementing different steps in the above methods through the creation module 610, acquisition module 620, determination module 630, and execution module 640, all the functions of the above device can be realized.

[0275] Furthermore, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments above, which will not be repeated here.

[0276] In this embodiment of the application, when the device 600 is a software device, the device can be deployed in the on-board computing device or edge computing unit of a mobile feeding device (such as an intelligent mixing and feeding vehicle).

[0277] Optionally, for some sub-modules of the device 600 that involve global data processing or require the use of ultra-high computing power (such as the aforementioned AI large model), they can also be distributed and deployed on edge servers in the farm or cloud computing device clusters, and work together with the mobile feeding device to complete the feed feeding task through the "end-edge-cloud" collaborative architecture.

[0278] For example, the aforementioned large AI model can be provided to farm users by cloud service providers as a cloud service (such as AI inference service) within a cloud management platform.

[0279] For example, a cloud management platform is primarily used to manage the infrastructure for running large AI models, which includes multiple data centers located in different regions, providing basic computing and storage resources for long-term historical data analysis and cross-sectoral macro-strategy reasoning.

[0280] Specifically, in actual feeding operations, the device 600 deployed in the mobile feeding equipment can automatically upload on-site sensing data (such as images, temperature and humidity) and input information such as the current remaining material quantity of the equipment, gathered through a soft bus near-field network, to the cloud environment via a wireless network (such as 4G / 5G) and an application programming interface. The AI ​​large-scale model strategy module in the cloud environment performs deep multimodal inference calculations to obtain an AI-recommended target feeding strategy (or a set of AI strategy parameters). Subsequently, this target feeding strategy is returned to the requesting mobile feeding equipment (i.e., the execution hardware) through the cloud environment. Upon receiving the strategy, the mobile feeding equipment drives its own electromechanical actuator to complete a precise feeding action within the target shed.

[0281] The above-mentioned "end-cloud" collaborative deployment method not only preserves the immediacy of local soft bus networking of mobile feeding equipment, but also improves the level of policy adaptation in abnormal scenarios by leveraging the massive computing power of the cloud.

[0282] The method provided in this application can be executed by a computing device. In the context of this application, the computing device can specifically be an onboard computing device or main control unit deployed inside a mobile feeding device (such as an intelligent mixing and feeding vehicle), a control module inside a loader or sensing device, or an edge server deployed locally at the farm or a remote cloud server (e.g., a computing device used to run AI large-scale model strategy modules). The hardware layer of the computing device includes processing units, memory, etc., and the operating system layer can run a real-time operating system or an industrial-grade operating system that supports a soft bus communication protocol. The execution entity of the method provided in this application can be the aforementioned computing device, or a functional module within the computing device capable of calling programs and executing specific business functions (such as automatic network discovery, feed list generation, feeding strategy calculation, etc.).

[0283] The following is combined Figure 7 This application provides a detailed description of a computing device provided in an embodiment.

[0284] Figure 7This is a schematic diagram of the architecture of a computing device 700 provided in an embodiment of this application. In practical applications, the control module inside the mobile feeding device, loading machine, and sensing device in this embodiment, or the edge / cloud server performing AI inference, can all adopt... Figure 7 The hardware architecture shown implements its underlying hardware support. Figure 7 The computing device 700 shown includes at least one processor 710 and a memory 720.

[0285] It should be understood that this application does not limit the number of processors and memories in the computing device 700.

[0286] The processor 710 executes instructions in the memory 720, causing the computing device 700 to implement the methods provided in this application (such as soft bus broadcast discovery, field sensing data acquisition, dynamic generation of feeding strategies, etc.). Alternatively, the processor 710 executes instructions in the memory 720, causing the computing device 700 to implement the various functional modules provided in this application, thereby implementing the methods provided in this application.

[0287] Optionally, the computing device 700 also includes a communication interface 730. The communication interface 730 uses a transceiver module, such as, but not limited to, a network interface card or a transceiver, to enable communication between the computing device 700 and other devices or communication networks.

[0288] Specifically, when the computing device 700 is used in a mobile feeding device, a loading machine, or a sensing device, the communication interface 730 can support short-range wireless communication technologies (such as Wi-Fi and Bluetooth) to enable near-field dynamic networking and data transmission functions of the soft bus communication module; it can also support cellular networks (such as 4G / 5G) to enable remote interaction with cloud management platforms or edge computing clusters.

[0289] Optionally, the computing device 700 also includes a system bus 740, wherein the processor 710, memory 720, and communication interface 730 are respectively connected to the system bus 740. The processor 710 can access the memory 720 through the system bus 740; for example, the processor 710 can perform data read / write or code execution in the memory 720 through the system bus 740. The system bus 740 can be a peripheral component interconnect express (PCIe) bus, an extended industry standard architecture (EISA) bus, or an industrial-grade bus architecture.

[0290] One possible implementation is that the processor 710 may be an integrated circuit chip with signal processing capabilities. By way of example and not limitation, the processor 710 can be a general-purpose processor (such as a central processing unit (CPU)), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices. In mobile feeding equipment, the processor 710 is typically an industrial-grade or automotive-grade chip with high reliability.

[0291] The memory 720 provides running space for processes in the computing device 700. By way of example and not limitation, the memory 720 is volatile memory (such as random access memory (RAM)) or non-volatile memory (such as read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), or flash memory).

[0292] For example, the memory 720 stores a creation module 610, an acquisition module 620, a determination module 630, and an execution module 640, so that the processor 710 can implement the feed feeding method for mobile feeding equipment provided in this application embodiment by calling and executing the program code of these modules.

[0293] In this embodiment, a computer program product containing instructions is also provided. When it runs on a computing device, it causes the computing device to execute the feed feeding method applied to the mobile feeding device in the above-described embodiments of this application, or causes the computing device to realize the function of the device 600 provided above.

[0294] In this embodiment, a computer-readable storage medium is also provided, which includes instructions. When the instructions in the computer-readable storage medium are executed on a computing device (such as the main control unit of a mobile feeding device or a cloud server), the computing device performs the intelligent aquaculture method based on a soft bus provided in the above-described embodiment.

[0295] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0296] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0297] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0298] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0299] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0300] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0301] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0302] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for intelligent aquaculture based on a soft bus, characterized in that, The method is applied to a mobile feeding device, which is equipped with a soft bus communication module. The method includes: Before the mobile feeding equipment enters the target shed, a second near-field network is automatically established with the target feeder through the soft bus communication module within the near-field range of the feed warehouse. The target feeder is equipped with a soft bus communication module. The second near-field network sends a feeding control command to the target feeder, so that the target feeder feeds the mobile feeding device according to the feeding list; After the mobile feeding device completes the feed dispensing, within the near-field range of the target shed, a first near-field network is automatically established with the sensing device deployed in the target shed via the soft bus communication module. The sensing device is equipped with a soft bus communication module. The on-site sensing data collected by the sensing device is obtained through the first near-field network; Based on the on-site sensing data and the preset feeding plan, a target feeding strategy is determined for the target shed. The feeding task is performed according to the target feeding strategy.

2. The method according to claim 1, characterized in that, The target feeding strategy includes at least one of the following: target single feeding amount, target feeding frequency, target discharge speed, and target execution action.

3. The method according to claim 1 or 2, characterized in that, The step of determining the target feeding strategy for the target shed based on the on-site sensing data and the preset feeding plan includes: The input dataset is determined based on the field sensing data and the feeding plan, wherein the input dataset includes the planned single feeding amount, the planned feeding frequency, the field status data of the target shed, and the remaining amount of material of the mobile feeding device; The target feeding strategy is determined based on the input dataset.

4. The method according to claim 3, characterized in that, Determining the target feeding strategy based on the input dataset includes: The target feeding strategy is calculated based on the input dataset and the preset feeding strategy formula.

5. The method according to claim 4, characterized in that, The step of calculating the target feeding strategy based on the input dataset and the preset feeding strategy formula includes: Using correction coefficients determined based on the on-site status data of the target shed, the planned single feeding amount and / or the planned feeding frequency are corrected to obtain the target single feeding amount and the target feeding frequency; The target discharge rate is calculated based on the target single feeding amount and the planned feeding time in the feeding plan.

6. The method according to claim 3, characterized in that, Determining the target feeding strategy based on the input dataset includes: Based on the input dataset, an artificial intelligence (AI) big data model is invoked to obtain the target feeding strategy, wherein the input information of the AI ​​big data model includes the input dataset, and the output information of the AI ​​big data model includes the target feeding strategy.

7. The method according to claim 3, characterized in that, Determining the target feeding strategy based on the input dataset includes: The first feeding strategy is calculated based on the input dataset and the preset feeding strategy formula. The second feeding strategy is obtained by calling the large AI model based on the input dataset. The first feeding strategy and the second feeding strategy are fused to obtain the target feeding strategy.

8. The method according to claim 7, characterized in that, The step of fusing the first feeding strategy and the second feeding strategy to obtain the target feeding strategy includes: Calculate the deviation between the first feeding strategy and the second feeding strategy in terms of control parameters and / or execution actions; Based on the deviation value, the target feeding strategy is determined using the corresponding fusion rule.

9. The method according to claim 8, characterized in that, The step of determining the target feeding strategy based on the deviation value using the corresponding fusion rule includes: When the deviation value is less than a preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent, the second feeding strategy is determined as the target feeding strategy; or When the deviation value is greater than or equal to a preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent, the control parameters in the first feeding strategy and the second feeding strategy are weighted and fused to obtain the target feeding strategy; or When the execution actions of the first feeding strategy and the second feeding strategy are inconsistent, an alarm is triggered to obtain the target feeding strategy for manual confirmation.

10. The method according to claim 1 or 2, characterized in that, The sensing device includes a monitoring camera and an environmental sensor, and the on-site sensing data includes at least one of the following: temperature or humidity collected by the environmental sensor, images of feed troughs, images of livestock feeding, and images of shed passages collected by the monitoring camera.

11. A smart aquaculture system based on a soft bus, characterized in that, The system includes a mobile feeding device, a target feeder deployed in a feed warehouse, and sensing devices deployed in a target shed. Each of the mobile feeding device, the target feeder, and the sensing devices is equipped with a soft bus communication module. The mobile feeding device is used to automatically establish a second near-field network with the target feeder through the soft bus communication module within the near-field range of the feed warehouse before entering the target shed, and to send feeding control commands to the target feeder through the second near-field network; The target feeder is used to add feed to the mobile feeding device according to the feed list in response to the feed control command; The mobile feeding device is also used to automatically establish a first near-field network with the sensing device through the soft bus communication module within the near-field range of the target shed after the feed is dispensed. The sensing device is used to collect on-site sensing data and send the on-site sensing data to the mobile feeding device through the first near-field network; The mobile feeding device is also used to determine a target feeding strategy for the target shed based on the on-site sensing data and the preset feeding plan. The mobile feeding device is also used to perform the feeding task of the feed according to the target feeding strategy.

12. The system according to claim 11, characterized in that, The target feeding strategy includes at least one of the following: target single feeding amount, target feeding frequency, target discharge speed, and target execution action.

13. The system according to claim 11 or 12, characterized in that, The mobile feeding device is specifically used for: The input dataset is determined based on the field sensing data and the feeding plan, wherein the input dataset includes the planned single feeding amount, the planned feeding frequency, the field status data of the target shed, and the remaining amount of material of the mobile feeding device; The target feeding strategy is determined based on the input dataset.

14. The system according to claim 13, characterized in that, The mobile feeding device is specifically used for: The target feeding strategy is calculated based on the input dataset and the preset feeding strategy formula.

15. The system according to claim 14, characterized in that, The mobile feeding device is specifically used for: Using correction coefficients determined based on the on-site status data of the target shed, the planned single feeding amount and / or the planned feeding frequency are corrected to obtain the target single feeding amount and the target feeding frequency; The target discharge rate is calculated based on the target single feeding amount and the planned feeding time in the feeding plan.

16. The system according to claim 13, characterized in that, The mobile feeding device is specifically used for: The target feeding strategy is obtained by calling an artificial intelligence (AI) big data model based on the input dataset, wherein the input information of the AI ​​big data model includes the input dataset, and the output information of the AI ​​big data model includes the target feeding strategy.

17. The system according to claim 13, characterized in that, The mobile feeding device is specifically used for: The first feeding strategy is calculated based on the input dataset and the preset feeding strategy formula. The second feeding strategy is obtained by calling the large AI model based on the input dataset. The first feeding strategy and the second feeding strategy are fused to obtain the target feeding strategy.

18. The system according to claim 17, characterized in that, The mobile feeding device is specifically used for: Calculate the deviation between the first feeding strategy and the second feeding strategy in terms of control parameters and / or execution actions; Based on the deviation value, the target feeding strategy is determined using the corresponding fusion rule.

19. The system according to claim 18, characterized in that, The mobile feeding device is specifically used for: When the deviation value is less than a preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent, the second feeding strategy is determined as the target feeding strategy. or When the deviation value is greater than or equal to a preset threshold and the execution actions of the first feeding strategy and the second feeding strategy are consistent, the control parameters in the first feeding strategy and the second feeding strategy are weighted and fused to obtain the target feeding strategy. or When the execution actions of the first feeding strategy and the second feeding strategy are inconsistent, an alarm is triggered to obtain the target feeding strategy for manual confirmation.

20. The system according to claim 11 or 12, characterized in that, The sensing device includes a monitoring camera and an environmental sensor, and the on-site sensing data includes at least one of the following: temperature or humidity collected by the environmental sensor, images of feed troughs, images of livestock feeding, and images of shed passages collected by the monitoring camera.

21. A device for intelligent aquaculture based on a soft bus, characterized in that, The device is deployed in a mobile feeding device, which also includes a soft bus communication module. The device comprises: A module is created to automatically establish a second near-field network with the target feeder within the near-field range of the feed warehouse before the mobile feeding equipment enters the target shed, wherein the target feeder is equipped with a soft bus communication module. The acquisition module is used to send a feeding control command to the target feeder through the second near-field network, so that the target feeder feeds the mobile feeding device according to the feeding list; The creation module is also used to automatically establish a first near-field network with the sensing device deployed in the target shed through the soft bus communication module in the near-field range of the target shed after the mobile feeding device completes the feed dispensing. The sensing device is equipped with a soft bus communication module. The acquisition module is also used to acquire the field sensing data collected by the sensing device through the first near-field network; The determination module is used to determine the target feeding strategy for the target shed based on the on-site sensing data and the preset feeding plan. An execution module is used to perform the feeding task of the feed according to the target feeding strategy.

22. A computing device, characterized in that, It includes a processor and a memory; the processor is configured to execute instructions stored in the memory to cause the computing device to perform the method as described in any one of claims 1 to 10.

23. A computer program product containing instructions, characterized in that, When the instructions are executed by the computing device, the computing device performs the method as described in any one of claims 1 to 10.

24. A computer-readable storage medium, characterized in that, It includes computer program instructions, which, when executed by a computing device, cause the computing device to perform the method as described in any one of claims 1 to 10.

Images