Delivery product processing method and device, computer equipment and storage medium

Abstract

The invention relates to a delivered product processing method and device, computer equipment and a storage medium. The method comprises the steps of determining a target processing module corresponding to a delivery scene where a delivered product is located according to acquired delivery processing information of the delivered product; for each target processing module, inputting the obtained input data of the target processing module into the target processing module; performing semantic understanding on the input data through the target processing module to obtain at least one processing task; each processing task is allocated to a corresponding processing unit for execution through the target processing module, and an execution result corresponding to each processing task is obtained; and determining a processing result corresponding to the target processing module through the target processing module according to the execution result corresponding to each processing task. By adopting the method and the device, different product delivery scenes can be adapted, and the manual processing cost of the delivered product is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial products, and in particular to a product delivery processing method and device, computer equipment and a storage medium. BACKGROUND

[0002] Currently, after the production of an industrial product, it needs to be delivered to a purchaser manually.

[0003] In the delivery link, a manual delivery personnel needs to provide the purchaser with a usage instruction, parameter adjustment and a solution for generating a landing application for the industrial product.

[0004] Currently, this manual delivery method requires a high labor cost, and the delivery personnel needs to have a high delivery skill. If the purchaser needs to be migrated to a different scene, the delivery personnel needs to understand the technical solution of the migration scene, and there is a problem of high migration cost and great difficulty. SUMMARY

[0005] Therefore, it is necessary to provide a product delivery processing method and device, computer equipment, computer readable storage medium and computer program product to adapt to different product delivery scenes and reduce the labor cost of product delivery.

[0006] In a first aspect, the present application provides a product delivery processing method, comprising:

[0007] According to the obtained delivery processing information of the delivery product, a target processing module corresponding to a delivery scene in which the delivery product is located is determined; the target processing module corresponding to the delivery scene includes at least one of the following: a solution planning module, a field analysis module and a training optimization module;

[0008] For each target processing module, the input data of the target processing module is input into the target processing module;

[0009] The target processing module performs semantic understanding on the input data to obtain at least one processing task;

[0010] The target processing module distributes each processing task to a corresponding processing unit for execution to obtain an execution result corresponding to each processing task;

[0011] The target processing module determines a processing result corresponding to the target processing module according to the execution result corresponding to each processing task.

[0012] In a second aspect, the present application provides a product delivery processing device, comprising:

[0013] The demand understanding module is configured to determine a target processing module corresponding to a delivery scenario in which the delivery product is located according to the obtained delivery processing information of the delivery product; the target processing module corresponding to the delivery scenario includes at least one of the following: a scheme planning module, a scene analysis module, and a training optimization module;

[0014] The data input module is configured to input the obtained input data of the target processing module into the target processing module for each target processing module.

[0015] The task processing module is configured to perform semantic understanding on the input data to obtain at least one processing task; distribute each processing task to a corresponding processing unit for execution to obtain an execution result corresponding to each processing task; and determine a processing result corresponding to the target processing module according to the execution result corresponding to each processing task.

[0016] In a third aspect, the present application provides a computer device, which includes a memory and a processor, the memory stores a computer program, and the processor implements the steps in the above method when executing the computer program.

[0017] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps in the above method when executed by a processor.

[0018] In a fifth aspect, the present application provides a computer program product, which includes a computer program, and the computer program implements the steps in the above method when executed by a processor.

[0019] The above delivery product processing method and device, computer device, computer readable storage medium, and computer program product can determine a target processing module corresponding to a delivery scenario, determine input data of the corresponding target processing module according to delivery processing information, divide the delivery scenario, and process the delivery scenario using a target processing module suitable for the delivery scenario, thereby improving the adaptability between processing functions and delivery scenarios for different products and different scene requirements. The target processing module performs semantic understanding on the input data to generate at least one processing task, distributes each processing task to a processing unit with different functions for processing to obtain an execution result corresponding to each processing task, and obtains a processing result corresponding to the target processing module, thereby dividing the entire delivery process into multiple stages, reducing the complexity of the delivery task, improving the flexibility and processing accuracy of the stage task, and solving the problem that the delivery personnel in the prior art need to have high delivery ability and technical understanding ability for different scenarios, thereby reducing the manual intervention in the delivery process and reducing the manual processing cost of the delivery product. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 An application environment diagram of a product delivery processing method provided by an embodiment of the present application is provided;

[0021] Figure 2 A flow diagram of a product delivery processing method provided by an embodiment of the present application is provided;

[0022] Figure 3 A structure diagram of a processing step provided by an embodiment of the present application is provided;

[0023] Figure 4 A structure diagram of a processing module provided by an embodiment of the present application is provided;

[0024] Figure 5 A structure diagram of a product delivery processing system provided by an embodiment of the present application is provided;

[0025] Figure 6 An application diagram of a processing module provided by an embodiment of the present application is provided;

[0026] Figure 7 A structure diagram of a product delivery processing apparatus provided by an embodiment of the present application is provided;

[0027] Figure 8 An internal structure diagram of a computer device provided by an embodiment of the present application is provided;

[0028] Figure 9 An internal structure diagram of another computer device provided by an embodiment of the present application is provided;

[0029] Figure 10 An internal structure diagram of a computer readable storage medium provided by an embodiment of the present application is provided. DETAILED DESCRIPTION

[0030] In order to make the objects, technical solutions and advantages of the present application clearer, further detailed description will be made to the present application with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0031] The product delivery processing method provided by the embodiments of the present application can be applied to, for example Figure 1The application environment shown. Among them, the terminal 102 communicates with the server 104 through the communication network. The data storage system can store the data required by the server 104 to process. The data storage system can be integrated on the server 104, or placed on the cloud or other network servers. Among them, the terminal 102 can be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, Internet of Things devices and portable wearable devices, Internet of Things devices can be smart speakers, smart televisions, smart air conditioners, smart vehicle devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented by an independent server or a server cluster composed of multiple servers.

[0032] As shown in Figure 2 The embodiments of the present application provide a processing method of delivering products. The method is applied to the terminal 102 or the server 104 in Figure 1 It can be understood that the computer device can include at least one of the terminal and the server. The method includes the following steps:

[0033] S201, according to the delivery processing information of the delivery product obtained, determine the target processing module corresponding to the delivery scene where the delivery product is located; The target processing module corresponding to the delivery scene includes at least one of the following: scheme planning module, field analysis module and training optimization module.

[0034] Among them, the delivery product can refer to an industrial product that has been produced and delivered to the user. The delivery processing information can be information that can achieve a specific target by processing the delivery product. The delivery processing information is information generated through human-computer interaction. The delivery scene can refer to the delivery link where the delivery product is located, or the delivery stage where the delivery product is located, etc. The target processing module can be applied in the delivery scene, and the delivery product is processed in the delivery scene to achieve a specific target.

[0035] In some embodiments, a complete delivery link of a delivery product can be divided into three general delivery stages through analysis of the complete delivery link. The three delivery stages can include, in order of execution: a whole scheme planning stage, a field data analysis stage, and a model training optimization stage. An interaction scenario can correspond to at least one delivery stage. For each delivery stage, a corresponding configuration module is configured. The target processing modules corresponding to the three delivery stages are: a scheme planning module, a field analysis module, and a training optimization module. The scheme planning module is applied in the whole scheme planning stage, and is used to construct a scheme for applying the delivery product. The field analysis module is applied in the field data analysis stage, and is used to analyze and process field data generated by the constructed scheme. The training optimization module is applied in the model training optimization stage, and is used to iteratively optimize the scheme for applying the delivery product.

[0036] In fact, the 0-1 complete delivery scenario includes the scheme planning module, the field analysis module, and the training optimization module, i.e., the three modules need to be executed in series. In this delivery scenario, there is no project scheme, and the scheme planning module needs to be initialized to generate a scheme, and then the field analysis stage and the training optimization stage exist. When subsequent optimization is performed, a complete project scheme does not need to be constructed again, and only field optimization and iterative optimization are performed, and the two optimizations can be performed in parallel. In some embodiments, if the training optimization is not performed, only the field analysis module can be run when the delivery link is in the middle of the project. If the user needs to perform model training optimization at a certain time, the field analysis module and the training optimization module can be run.

[0037] S202, for each target processing module, input the obtained input data of the target processing module into the target processing module.

[0038] The content and data format of the processing data required by different target processing modules can be different. The input data satisfying the required content and data format can be determined according to the delivery processing information and the content and data format required by the target processing module.

[0039] In some implementations, when the number of delivery scenarios is one, the number of target processing modules is one accordingly, and the input data can be directly input to the target processing module to obtain the processing result corresponding to the target processing module. When the number of delivery scenarios is at least two, the number of target processing modules is at least two, and each target processing module needs to be executed in sequence. At this time, the input data of the target processing module is determined according to the delivery processing information, and the execution result of the target processing module executed in the previous order and relied on by the target processing module is taken as part of the input data in the target processing module. When the target processing module is the first executed target processing module, the input data of the target processing module can be directly determined according to the delivery processing information.

[0040] S203, performing semantic understanding on the input data by the target processing module to obtain at least one processing task.

[0041] Among them, one target processing module can have the functions of task generation and distribution, task execution, and result integration. In some embodiments, the target processing module also has the function of correctness verification. The semantic understanding of the input data can refer to determining the target to be completed from the input data, and for each target, a processing task can be generated accordingly. The processing task can be further divided into subtasks by the target processing module to be executed.

[0042] S204, distributing each processing task to the corresponding processing unit for execution by the target processing module to obtain the execution result corresponding to each processing task.

[0043] Among them, the target processing module is configured with at least one processing unit. Each processing unit is used to execute at least one processing task. Among them, the processing unit can be a large model agent. The large model can encode the input data to obtain a feature vector, and decode the feature vector to obtain output data. The large model agent can be a system with autonomous perception of the environment, dynamic decision-making and active execution capability, which can achieve complex goals through tool calling and environment interaction. The large model agent uses a large model as a tool, and combines a planning module, a memory system and a tool library to form a complete action chain of "perception→decision→execution→optimization", and achieve complex goals.

[0044] In some embodiments, the processing unit is configured to invoke the large model to perform the assigned at least one processing task. The input data of each processing task can be determined according to the input data of the target processing module, and the input data of the processing task is processed according to the content of the processing task to generate input data that can be understood by the large model, and the input data is provided to the large model to obtain the output data of the large model. Wherein, the input data of the processing task can be added to the prompt template to obtain the input data that can be understood by the large model.

[0045] According to the specific content of the processing task, the output data can be taken as the execution result corresponding to the processing task, or the corresponding operation can be further performed according to the output data to obtain the execution result corresponding to the processing task, for example, according to the output data to determine the called application and the parameter, to execute the application by providing the parameter to the application, and to call the application to obtain the response result of the application, and to determine the response result as the execution result corresponding to the processing task.

[0046] In some embodiments, each processing task can be executed in parallel or in series. The processing task can depend on the execution result corresponding to the other processing task, and accordingly, the execution result corresponding to one processing task can be taken as the input data of the processing task that depends on the execution result.

[0047] S205, determining the processing result corresponding to the target processing module according to the execution result corresponding to each processing task through the target processing module.

[0048] In some embodiments, the execution results corresponding to each processing task can be aggregated to obtain the processing result corresponding to the target processing module. The execution results corresponding to each processing task can also be subjected to correctness verification, and when the correctness verification passes, the execution results corresponding to each processing task are aggregated to obtain the processing result corresponding to the target processing module. The execution result corresponding to the last executed processing unit in the target processing module can be determined as the execution result corresponding to the target processing module.

[0049] It can be seen that in the embodiment of the present application, by determining the target processing module corresponding to the delivery scenario and determining the input data of the corresponding target processing module according to the delivery processing information, the delivery scenario is divided and processed using a target processing module that is adapted to the delivery scenario. The adaptability between the processing function and the delivery scenario can be improved according to the requirements of different products and different scenarios. The input data is semantically understood by the target processing module to generate at least one processing task, and each processing task is assigned to a processing unit with different functions for processing. The execution result corresponding to each processing task is obtained, and the processing result corresponding to the target processing module is summarized to obtain the processing result corresponding to the target processing module, thereby splitting the entire delivery link into multi-stage processing, which can reduce the complexity of the delivery task, improve the flexibility and processing accuracy of the staged tasks, and solve the problem in the prior art that the manual delivery process requires the delivery personnel to have high delivery capabilities and technical understanding capabilities of different scenarios, resulting in delivery labor costs. It can reduce manual intervention in the delivery link and reduce the manual processing cost of the delivered product.

[0050] In some embodiments, the target processing module allocates each processing task to a corresponding processing unit for execution, and obtains the execution result corresponding to each processing task, including:

[0051] Allocate each processing task to the corresponding processing unit through the target processing module;

[0052] The processing unit to which the processing task is assigned determines the processing steps corresponding to the processing function according to the processing function required by the processing task; the processing unit is configured with at least one processing function, and each processing function is configured with at least one corresponding processing step;

[0053] The processing unit to which the processing task is allocated processes the processing task according to the processing steps corresponding to the processing function to obtain the execution result corresponding to the processing task.

[0054] Among them, the processing unit has different processing functions and can perform different processing tasks. The processing function can refer to a behavior pattern (characters), and each behavior pattern has its own objective corresponding processing steps (steps). A processing function can correspond to at least one processing step. Different processing functions correspond to different processing steps. In fact, the functions that a processing unit can achieve are guided to a certain extent by the steps of the processing function. The breadth and rigor of the processing step guidance will bring different effects. Processing the processing task according to the processing steps corresponding to the processing function can be to add the processing steps to the prompt template so that the large model can process the processing task according to the processing steps.

[0055] The processing function is configured with a processing step, which can finely control and guide the differences in the operations performed by the processing unit, so that the delivered product can quickly adapt to different application scenarios to provide the required functions. Moreover, there is no need to switch application scenarios by maintaining code, which can reduce the migration cost of the agent, for example, switching from an industry-oriented task to a financial analysis scenario.

[0056] The processing unit can call a large model, and the processing function can be understood as the role of the large model. By explicitly defining the identity, behavior boundary and output style of the model in the interaction through the prompt template, the model simulates specific professions, scenarios or character features, which essentially constrains the generation logic and expression method of the model. The processing step can be understood as a more fine-grained constraint on the generation logic and expression method of the model.

[0057] Each target processing module has multiple processing units responsible for multiple behavior patterns, and each behavior pattern corresponds to at least one processing step. The characteristics of each processing unit and the processing steps that need to be considered for each characteristic are different. For example, a delivery project team may have PM (Project Manager), software management, and algorithm management personnel, each responsible for a different range of work, and there may be multiple algorithms, such as an algorithm management personnel responsible for traditional algorithms and another algorithm management personnel responsible for deep learning algorithms, which is equivalent to two working ranges for algorithm management personnel, i.e., two behavior patterns for a processing unit, because there are differences between traditional algorithms and deep learning algorithms, the design process of the two behavior patterns, i.e., the processing steps, will be different. Therefore, different behavior patterns have different processing steps. In other words, the things a processing unit can do are guided to some extent by the processing steps of the behavior pattern, and the extent of the guidance will result in different effects, such as some people in the same profession being more conservative and some being more spontaneous.

[0058] In some implementations, when a large model is called, processing functions and processing steps are added in the prompt template and provided to the large model, which implements the processing functions and processing steps as inputs to the large model, so that when the large model performs a processing task, the processing logic corresponding to the processing functions and processing steps can be implemented to process the execution task according to the constrained content.

[0059] The embodiments of the present application configure corresponding component supports for the scheme planning module, the field analysis module, and the training optimization module, and complete the overall interaction through a global configuration protocol, providing a feasible implementation for the introduction of general artificial intelligence technology.

[0060] It can be seen that the processing unit to which the processing task is assigned determines the corresponding processing step according to the processing function required by the processing task, processes the processing task according to the processing step corresponding to the processing function, and obtains the execution result corresponding to the processing task. The fine-grained difference operation executed by the processing unit can be controlled and guided, so that the processing unit can quickly switch between different application scenarios, realize the delivery product quickly adapted to different application scenarios, provide the required function, reduce the migration cost and complexity of the agent.

[0061] In some embodiments, the processing unit to which the processing task is assigned processes the processing task according to the processing step corresponding to the processing function, and obtains the execution result corresponding to the processing task, including:

[0062] The processing unit to which the processing task is assigned executes in turn according to the execution order of the processing step corresponding to the processing function;

[0063] When the target processing step is executed, the execution result corresponding to the target processing step is generated according to the input data and the processing reference information of the target processing step;

[0064] According to the verification reference information of the target processing step, the execution result corresponding to the target processing step is verified to obtain a verification result;

[0065] When the verification result is verified, the execution result corresponding to the processing task is determined according to the execution result corresponding to the target processing step.

[0066] Each processing step is configured with details, and has clear input and output requirements. The processing reference information is used to constrain the processing method, constrain the input and constrain the output, etc. The verification reference information is used to verify the correctness of the execution result corresponding to the processing step.

[0067] Among them, for each processing step, the execution result corresponding to the processing step can be used as the input data of the dependent next processing step, or the execution results corresponding to other processing steps are summarized to obtain the execution result corresponding to the processing function, and the execution results corresponding to each processing function are summarized to obtain the execution result corresponding to the processing unit. When the processing task assigned to the processing unit is only implemented by one processing function, the execution result corresponding to the processing function can be determined as the execution result corresponding to the processing unit.

[0068] In one example, the input data of a target processing step is data to be analyzed, the target processing step processes to obtain output data conforming to processing reference information, and the output data is checked again according to the check reference information, such as whether there is a missing key field constrained by the processing reference information. In some implementations, the analyst will integrate the data, count the data according to the database format template, which is equivalent to being constrained by the processing reference information. The next step is to obtain the data counted according to the format template to perform related tasks, and after step-by-step disassembly, the equivalent of sequentially executed multiple processing steps is obtained. Thus, many routine tasks can be divided into related task steps, and can be correctly converted and unified with the actual processing process.

[0069] As shown in Figure 3 A processing function includes four processing steps (stepi, i is 1-4). Each processing step is configured with processing reference information and check reference information, wherein the processing reference information defines input, output, notes, and additional special prompts. The check reference information defines inspection items. Each processing step first confirms the input dependency, obtains the input data, which can include the input data of the processing unit and the output of other processing steps. For example, the input dependency of processing step 2 (step2) includes the output (S1) of processing step 1 (step1). The processing step generates output according to the input data, the notes and the regulations, that is, according to the processing reference information, to obtain the execution result (Si, i is 1-4) corresponding to the processing step. Finally, the processing step checks the output format and the regulations, that is, the correctness of the execution result is checked according to the check reference information.

[0070] It can be seen that by configuring each target processing step with processing reference information to constrain the input and output of the target processing step, and configuring the target processing step with check reference information to constrain the correctness checking process of the target processing step, the target processing step is controlled and planned in a fine-grained manner, the content of the target processing step can be flexibly adjusted to adapt to more complex tasks, and for each processing module behind the processing unit, a step-by-step task table is introduced, each processing unit is given different functions to cope with different work, and the step-by-step task table is used to configure the thinking steps related to the functions and the event dimensions that need to be focused on for each processing unit. The framework for building general artificial intelligence agents can quickly switch between function states, and when facing different non-creative tasks, it can quickly switch and be guided and regulated by human experience.

[0071] In some embodiments, the semantic understanding of the input data by the target processing module obtains at least one processing task, including: the semantic understanding of the input data by the planning unit obtains at least one processing task; and the distribution of each processing task to a corresponding processing unit by the target processing module, including: the distribution of each processing task to a corresponding processing unit by the planning unit. After the processing task is processed by the processing unit to which the processing task is distributed according to the processing step corresponding to the processing function, the execution result corresponding to the processing task is obtained. The method further includes: the verification of the execution result corresponding to each processing unit by the verification unit according to the verification reference information of each processing unit; when the execution result corresponding to each processing unit is verified, the execution result corresponding to each processing unit is provided to the integration unit; when the execution result corresponding to the processing unit is not verified, the interaction with the processing unit that does not pass is performed to make the processing unit that does not pass rethink the answer, update the execution result, and verify the updated execution result again until the updated execution result passes or the number of processing times of the processing unit that does not pass is greater than or equal to a preset number threshold (such as 3 times), and a warning is given; and the integration of the execution result of each processing unit that passes the verification by the integration unit, and the output according to the module reference information.

[0072] As Figure 4As shown, generally one processing module (standard Module Unit) includes a planning unit, a verification unit, a processing unit and an integration unit by default. Each unit performs a user delivery behavior simulation task. Among them, the planning unit is used to assign the relevant tasks to the corresponding unit (mainly the processing unit) for execution according to the demand generated by an event and according to the ability boundary of each unit, specifically to achieve task understanding and task splitting, obtain each processing task, and distribute each processing task to each processing unit (agenti, i represents the ith agent). The planning unit can summarize and refine the corresponding execution results obtained from history, i.e. context, and convert human description into language content understandable by the agent. The planning unit can determine the ability boundary and the demand that can be met (i.e. the processing function that can be implemented) of the processing unit according to the knowledge base. Different processing units correspond to different knowledge bases, and different knowledge bases have different processing capabilities and knowledge of different tasks. The data volume and content of the knowledge base of each processing unit are independent of each other. Each processing unit executes the assigned processing task according to the processing steps corresponding to the configured processing function (character) to obtain the execution result. The verification unit performs correctness verification on each processing task, i.e. the execution result corresponding to the processing unit, i.e. realizes self-checking of the processing unit. If there is a problem, it will communicate with the corresponding processing unit to make the processing unit think again about the execution result. For example, the correctness verification range of the verification unit on the processing unit in the scheme planning module includes: the format planning, task planning and configuration defined by the processing unit must exist verification.

[0073] In the planning unit, the demand of the last stage will be referred to confirm whether there is a related processing unit in the knowledge base and the ability range, and the processing task, demand and refined knowledge summary will be distributed to the corresponding processing unit. In the task execution stage, the feedback exception, demand, processing step and matching knowledge will be used as the input of the processing unit, and the final result will be completed by the processing function represented by each processing unit independently or cooperatively. Among them, the exception is the exception description of the error generated by the verification unit after verifying the output data of the processing unit. The exception is used to correct the second thinking of the processing unit on the processing task. In fact, for the same demand description content, i.e. for the same processing task, the processing unit may have multiple thoughts, and the processing unit combines the last execution result and the new rectification demand to perform new execution on the processing task.

[0074] It can be seen that by configuring the general planning unit, processing unit and verification unit, etc., the maintenance cost of the solution user in the industrial delivery can be reduced, and the training cost of the user using the delivery product can be reduced, the flow change of the production line personnel and the delivery personnel in the industrial delivery is greatly stabilized, and the overall stability and team building of the industrial delivery team is greatly improved.

[0075] In some embodiments, according to the obtained delivery processing information of the delivery product, the target processing module corresponding to the delivery scene where the delivery product is located is determined, including:

[0076] By the requirement understanding module, the semantic understanding of the obtained delivery processing information of the delivery product is performed according to the processing module configuration information, and the requirement description content corresponding to at least one target processing module is obtained.

[0077] For each target processing module, the requirement description content of the target processing module is determined as the input data of the target processing module.

[0078] The processing module configuration information is used to determine the function of the processing module. The requirement description content can be the requirement information that the target processing module can understand. The semantic understanding of the delivery processing information to obtain the requirement description content can be to reconstruct the delivery processing information into language content that the processing module can understand. The target processing module with the requirement description content is the processing module required to process the delivery processing information. The requirement description content can be the content that needs to be processed by the corresponding target processing module and is split from the delivery processing information. The requirement description content can be the description content of the task that needs to be processed by the corresponding target processing module, so that the requirement description content is determined as the input data of the target processing module, which can be to assign the task in the delivery processing information that needs to be executed by the target processing module to the target processing module for execution.

[0079] As Figure 5The processing system of the delivery product of the industrial scene delivery shown is executed according to the user's demand to realize the final result. The user's demand includes customer demand (the object of the delivery product is the customer), developer demand, pre-sale demand and other project demand, etc. Different industrial delivery scene configurations can be adapted. The demand understanding module understands the demand and task allocation of the delivery processing information according to the foregoing user demand and knowledge base, etc., and forms the demand description content of at least one processing module. Among them, the processing module can include a scheme planning module, a field analysis module and a training optimization module, etc. The execution result corresponding to each target processing module is executed by Function Call (function call). Through the function call as the execution result corresponding to the target processing module, different output sources can execute different function scheduling. For example, the scheme planning module outputs a configuration file, which needs to call the corresponding developed algorithm to parse the configuration file to realize the execution of the configuration file. For example, the output result of the field analysis module needs to call the related online statistical application to complete parameter updating, report output and automatic adjustment of overkill and missing, etc. In this way, the user's demand can be directly converted into the corresponding execution operation, reducing the user's use difficulty for the related application, reducing the training difficulty, making the processing operation of the delivery product closer to unmanned operation, converting from user learning operation to user expressing intention, letting the system automatically judge the intention target and execute the related operation, reducing the processing cost and difficulty of the delivery product. By means of the function call of the large model, the function layer and the configuration of the solution are separated, the specific function is aligned through the function description of the cfg level, so that the processing unit avoids the construction of the code, and the aligned module function can be equivalently effective through the cfg configuration, which well avoids the code illusion problem of the large model.

[0080] The demand understanding module is configured with a processing unit, and the processing unit is configured with a processing function. The input of the demand understanding module, that is, the input of the processing function, includes: project sop (standard operating procedure), user demand, data information, project parameter information, project knowledge base (agent default configuration, different agent knowledge bases are different), and self specification requirements, etc. Among them, the project sop can include detection requirements and technical agreement, etc. Some special demands of the user can include: hope to process fast or hope that the scheme involves low hardware cost. The data information can include picture data, label data and label specification sop, etc. The project parameter information can include camera, light source parameter and light source type, etc. The introduction of the project knowledge base makes the demand understanding module have more knowledge basis that can be referred to, for example, the knowledge of the project management user can be provided to the demand understanding module, so that the demand understanding module can disassemble the customer demand, establish the development demand and convert the pre-sale demand, etc., and then use the appropriate demand description content to transfer to the corresponding downstream processing module. The delivery processing information at least includes: user demand. The prompt includes self specification requirements. In an example, the prompt includes self specification requirements: 1, your output must include XX1, XX2 and XX3, etc., which need to consider xxx, for the station thinking of XX2, must output a, b, c and d, etc., under the condition of complying with the format specification and the matching condition of the related knowledge, you can make necessary additional supplement.

[0081] The processing function 1: Charater1 is configured with a processing step step, and the processing step is defined as: 1. checking whether the dimension of the input information is missing (assuming that 1, 2, 3 and 4 have the format template of the requirement standard); 2. demand understanding for the input; 3. disassembling the standard demand and the customer description, and matching and understanding the knowledge base for the customer description; 4. after completing the demand understanding, matching and distributing between the processing task and the processing unit.

[0082] The output of the demand understanding module, that is, the output of the processing function, includes: 1. the demand description content of the scheme planning module, that is, the task demand description of the scheme planning module. If the demand description content is too long, the demand understanding module appropriately reduces the content, or uses the output in the form of a table, and in addition to the output required in the foregoing, the processing unit or the processing module can be allowed to add output content under the condition of complying with the specification.

[0083] Specifically, the demand description content of the scheme planning module is:

[0084] 1. Project sop:

[0085] - missing kill requirement xxx, overkill requirement xxx, requirement key defect scratch not allowed to miss kill

[0086] - Need to support n types, type-size-parameter relationship

[0087] 2. Each station refinement information

[0088] Station 1,

[0089] a: use xxx scheme;

[0090] b: there are xx defects,

[0091] c: each defect detection requirement,

[0092] d: related picture list

[0093] Station 2,

[0094] a: use xx scheme

[0095] b: there are xx defects

[0096] c: each defect detection requirement

[0097] d: related picture list

[0098] 3. Demand summary

[0099] - Suggest using xxx small model

[0100] - The related model of key defect xxx needs to ensure the effect

[0101] According to the requirements, please design the related scheme

[0102] In addition, the requirement understanding module is also configured with a verification unit, which verifies the correctness of the output of the processing unit, for example, whether the content in the self-specification requirement exists or not.

[0103] As can be seen, by matching the delivery processing information and the processing module through the requirement understanding module, the target processing module is obtained, and the delivery processing information is converted into the requirement description content that the target processing module can understand, and is allocated to the corresponding target processing module, so that the target processing module executes the task indicated in the delivery processing information, which can automatically decompose complex requirements into sub-requirements and allocate them to the corresponding functional processing module for execution, dynamically plan the task chain, adapt to flexible and diverse requirements, and each sub-requirement can be executed independently, thereby improving the task processing efficiency, in addition, the sub-requirements can be executed specifically, which can improve the accuracy of the execution result, and the current solution framework is changed from how people learn to use to people describing user requirements with descriptive sentences, automatically understanding the meaning of the description, and executing related tasks, and finally helping people complete.

[0104] In some embodiments, each processing task is assigned to a corresponding processing unit for execution by the target processing module, obtaining an execution result corresponding to each processing task, including:

[0105] When the target processing module is a scheme planning module, a project understanding task is assigned to a project understanding unit for execution by the scheme planning module; the input data of the project understanding task is processed by the project understanding unit to obtain an execution result corresponding to the project understanding task; the input data of the project understanding task includes project standard process information and text information of a product station; the execution result corresponding to the project understanding task includes current scheme text description information, historical scheme text description information, and scheme configuration text description information;

[0106] The visual understanding task is assigned to a visual understanding unit for execution by the scheme planning module; the input data of the visual understanding task is processed by the visual understanding unit to obtain an execution result corresponding to the visual understanding task; the input data of the visual understanding task includes project standard process information and image information of a product station; the execution result corresponding to the visual understanding task includes current scheme image description information, historical scheme image description information, and scheme configuration image description information;

[0107] The scheme construction task is assigned to a scheme construction unit for execution by the scheme planning module; the input data of the scheme construction task is processed by the scheme construction unit to obtain an execution result corresponding to the scheme construction task; the input data of the scheme construction task includes the execution result corresponding to the project understanding task, the execution result corresponding to the visual understanding task, and user demand information; the execution result corresponding to the scheme construction task includes a delivery scheme.

[0108] The project understanding unit is configured to understand the text information. The visual understanding task is configured to understand the image information. The scheme construction unit is configured to generate a scheme based on the understanding result of the text information and the understanding result of the image information. The project standard process information is a general process of a project. The product station can be a station where a product is placed or used. The current scheme text description information can be the content of the scheme extracted from the input text information. The historical scheme text description information can refer to the content of the scheme accumulated from historical text information. The scheme configuration text description information can refer to the text configuration information of the required scheme. The current scheme image description information can be the content of the scheme extracted from the input image information. The historical scheme image description information can refer to the content of the scheme accumulated from historical image information. The scheme configuration image description information can refer to the image configuration information of the required scheme. The user demand information can be the demand content extracted from the delivery processing information and added to the demand description content of the scheme planning module.

[0109] In one example, the scheme planning module includes three processing units, a project understanding unit, a visual understanding unit and a scheme construction unit.

[0110] The input of the scheme planning module is the requirement description content of the output of the requirement understanding module. The planning unit of the scheme planning module performs basic checking and splitting, and the processing unit of the scheme planning module performs vertical focus tasks. Accordingly, the processing steps of the planning unit do not have specific summaries, and only checking and processing tasks are matched and distributed.

[0111] The project understanding unit has a processing function, and the input of the processing function is: 1. project sop and 2. each station refinement information. The processing function of the project understanding unit corresponds to the following processing steps: 1. input 1. project standard process information and 2. product station text information; 2. match the scheme with the knowledge base; 3. summarize the scheme suggestion of the project. Among them, the project standard process information is like the project sop, the product station text information is like the each station refinement information, and the input 1. project sop and 2. each station refinement information do not contain images. The output of the processing function is: 1. simplified and / or refined current scheme text description information and historical scheme text description information; 2. scheme configuration text description information. The scheme configuration text description information is like the configuration information (cfg) of each station.

[0112] The visual understanding unit has a processing function, and the input of the processing function is: 1. project sop and 2. each station refinement information. The processing function of the visual understanding unit corresponds to the following processing steps: 1. input 1. project standard process information and 2. product station image information; 2. match the scheme with the knowledge base; 3. further refine the station information design scheme and picture features; 4. match and analyze the picture features and the station information, and summarize the scheme in combination with the knowledge base. Among them, the product station image information is like the image associated with the each station refinement information, and the input 1. project sop and 2. each station refinement information contain images. The output of the processing function is: 1. simplified and / or refined current scheme image description information and historical scheme image description information; 2. scheme configuration image description information. The scheme configuration image description information is like the scheme configuration after understanding the picture features.

[0113] The scheme construction unit has two processing functions, and the input of processing function 1: 1. The execution result corresponding to the project understanding task, 2. The execution result corresponding to the visual understanding task and 3. The user demand information. Among them, the user demand information can be the demand summary output by the previous case demand understanding module. The processing steps defined for processing function 1 are: 1. Check if the input is missing; 2. Check if the workstations of input 1 and input 2 are complete; 3. Confirm the scheme differences of the picture features, and list 1, 2 and 3; 4. Further analyze the scheme differences, and match the knowledge base for summary; 5. Summarize and adjust each workstation scheme; 6. Confirm the customer demand summary, refer to the knowledge base, and adjust the parameters of each scheme accordingly; 7. Check the standard software driven cfg format. The output of processing function 1: 1. The cfg output of all schemes; 2. The demand test request.

[0114] The input of processing function 2: 1. The output of processing function 1 and 2. The user demand information. Among them, the user demand information can be the demand summary output by the previous case demand understanding module. The processing steps defined for processing function 2 are: 1. Split each workstation; 2. Call different function applications for testing and feedback; 3. Statistic whether the feedback meets the user's requirements; 4. Statistic whether the feedback test scheme can be successfully constructed. The output of processing function 2: deliver scheme. The execution result corresponding to processing function 2, that is, the output, is determined as the execution result corresponding to the scheme construction unit. In the scheme planning module, the delivery scheme is the final goal of the scheme planning module, so as to summarize the execution results corresponding to the project understanding unit, the visual understanding unit and the scheme construction unit, and obtain the execution result corresponding to the scheme planning module as the delivery scheme.

[0115] As can be seen, by configuring the scheme planning module to include the project understanding unit, the visual understanding unit and the scheme construction unit, and configuring the corresponding input, output and processing steps for each processing unit, the function of the scheme planning can be accurately matched according to the demand of the delivered product, and the controllability of the scheme planning is increased.

[0116] In some embodiments, by the target processing module, each processing task is allocated to the corresponding processing unit for execution to obtain the execution result corresponding to each processing task, including:

[0117] When the target processing module is the field analysis module, by the field analysis module, the field analysis task is allocated to the field analysis unit for execution; by the field analysis unit, the input data of the field analysis task is processed to obtain the execution result corresponding to the field analysis task; the input data of the field analysis task includes: current field detection result, historical detection result and corresponding user demand information, current user demand and user interaction feedback information; the execution result corresponding to the field analysis task includes: calling the application corresponding to the field analysis task and the calling result.

[0118] The field analysis unit is configured to analyze the field data. The current field detection result can be a result generated by a detection operation of the field user based on a current scene. The historical detection result can be a result generated by a detection operation based on a historical scene. The user demand information corresponds to the historical detection result, and the user demand information can be a user demand that can obtain the historical detection result. In fact, the user has a certain demand, and based on the demand, the user performs a corresponding detection operation in the historical scene to generate the historical detection result. The current user demand can be a demand of the user in the current scene. The user interaction feedback information can be feedback of the user to an execution result corresponding to the field analysis module executed multiple times in the current scene. In fact, the user needs to continuously debug the delivered product or the software and hardware system associated with the delivered product according to the execution result corresponding to the field analysis module, and each time of debugging generates one piece of data. The generated data can be understood as the user interaction feedback information of the user interacting with the execution result, and the field analysis module can perform rethinking processing on the updated data (i.e., the user interaction feedback information). In addition, the user can feed back to the execution result corresponding to the field analysis module, thereby generating the user interaction feedback information, and the field analysis module performs rethinking processing on the user interaction feedback information. The application corresponding to the field analysis task can be an application for implementing data analysis. The application corresponding to the field analysis task can be a function call. The calling result can be a response result obtained by calling the application corresponding to the field analysis task.

[0119] The input of the field analysis module includes: 1. detection results and / or pass / fail analysis results; 2. new demands of the user, such as shielding a certain defect (project level) or generating a report (software level); and 3. user interaction (pass or fail). The planning unit of the field analysis module performs semantic understanding according to the demand description content of the scheme planning module, and then generates and distributes processing tasks according to the knowledge base. The field analysis module is configured with one processing unit, i.e., a field analysis unit.

[0120] The input of the field analysis unit includes: 1. historical detection results and corresponding user demand information. The user demand information can be a report obtained according to the pass / fail analysis.

[0121] The field analysis unit has two processing functions. The input of the processing function 1 includes: 1. current field detection results. The processing steps corresponding to the processing function 1 are defined as: 1. result completeness field detection; 2. understanding the detection results by referring to the knowledge base and standard project protocols; 3. analyzing the customer demand and basic optimization targets (optimizing pass / fail) for analysis; and 4. matching the analyzed targets and related software protocol operations. The output of the processing function 1 includes: 1. problems to be optimized a and suggested operations for each problem; and 2. a summary report of the current field detection results.

[0122] The inputs of processing function 2 are: 1. the output of processing function 1, 2. the current user demand, and 3. user interaction feedback information. Among them, the current user demand can be all the images cropped by the control. The processing steps corresponding to processing function 2 are defined as follows: 1. Sort out user needs and feedback, and match the software operations; 2. Sort out the optimization problem b, and match the software operations; 3. Perform software testing and feedback on function call; 4. Perform a closed-loop check on problem a and problem b to see if they are completed and successful; 5. Check whether the effect has achieved the target improvement. If not, it will automatically trigger a self-check failure function call. The output of processing function 2: The output (function call level) directly calls the application and prompts the user to feedback whether it passes. If not, it will be analyzed again.

[0123] The execution result corresponding to the processing function 2 is determined as the execution result corresponding to the field analysis unit. In the field analysis scenario, calling through is the ultimate goal of the field analysis module.

[0124] like Figure 6 The detailed process of the on-site analysis module is shown below. First, the planning unit understands various requirements. Then, based on the relevant knowledge base A, it converts the received requirement descriptions into requirements that can be understood in the next phase. This planning unit actually reconstructs the requirements. This reconstruction phase is dependent on the accuracy of knowledge base matching, necessitating the additional configuration of a planning unit to perform requirement reconstruction. The planning unit then determines whether a processing unit with the relevant processing capabilities exists for the requirement, based on the requirements from the previous phase, in knowledge base B, and the capabilities of each processing unit. It then assigns the task, requirement, and refined knowledge summary. Alternatively, an additional task assignment unit can be configured. This unit, based on the requirements from the previous phase, in knowledge base C, and the capabilities of each processing unit agent, also determines whether a processing unit with the relevant processing capabilities exists for the requirement. It then assigns the task, requirement, and refined knowledge summary. During the task execution phase, the processing units process the feedback on anomalies, requirements, and the steps and associated knowledge that have been matched. The final results of the on-site analysis module are then independently or collaboratively completed by the processing functions represented by each processing unit.

[0125] It can be seen that by configuring the on-site analysis module including the on-site analysis unit and configuring the corresponding input, output and processing steps for the on-site analysis unit, the on-site data of the delivered product can be analyzed, and the call debugging can be continuously optimized and called according to user feedback. The call information of the delivered product can be flexibly adjusted to achieve the goal, thereby reducing the labor cost of on-site analysis and debugging of the delivered product.

[0126] In some embodiments, each processing task is assigned to a corresponding processing unit for execution by the target processing module, obtaining an execution result corresponding to each processing task, including:

[0127] When the target processing module is a training optimization module, the training optimization module assigns a solution generation task to a solution generation unit for execution; the solution generation unit processes the input data of the solution generation task to obtain an execution result corresponding to the solution generation task; the input data of the solution generation task includes: problem images, historical problems, and corresponding requirements and iteration indicators; the execution result corresponding to the solution generation task includes solution content and unprocessable problems;

[0128] The training optimization module assigns an optimization task to an optimization unit for execution; the optimization unit processes the input data of the optimization task to obtain an execution result corresponding to the optimization task; the input data of the optimization task includes: the execution result corresponding to the solution generation task, user demand information, and user feedback; the execution result corresponding to the optimization task includes problem alarms.

[0129] The solution generation unit is used to generate the content of the solution to the problem and determine the problem that cannot be solved. The optimization unit is used to solve the problem and alarm the problem that cannot be solved. The problem image can refer to the image associated with the existing problem. The historical problem can refer to the historical problem. The requirement corresponds to the historical problem, and the corresponding requirement can refer to the user demand that can solve the historical problem. In fact, the user has a certain demand, and based on the demand, the user can solve the historical problem in the historical scene. The iteration indicator can refer to the parameter or direction of iterative optimization, etc. The solution content can refer to the content of the solution to the problem. The unprocessable problem can refer to the problem for which no solution is provided or the problem that has not been found. The user feedback can refer to the user feedback on the execution result corresponding to the solution generation task. The user can feed back on the execution result corresponding to the solution generation task, thereby generating user interaction feedback information, and the training optimization module reconsiders the user interaction feedback information. The problem alarm can refer to the problem that cannot be solved finally.

[0130] The input of the training optimization module is: 1. Problem image; 2. Historical problem and corresponding requirement; 3. Iteration indicator; 4. User feedback; 5. User demand. The planning unit of the training optimization module performs semantic understanding according to the requirement description content of the training optimization module, and then generates and distributes processing tasks according to the knowledge base. The training optimization module is configured with two processing units, namely the solution generation unit and the optimization unit. The problem image can include the detection result of the scene and / or the summary of the kill statistics. The user demand can include adding a new defect. The user feedback can include the image of the problem that still exists.

[0131] The solution generation unit has two processing functions, and the input of processing function 1 is: 1. problem image and 2. historical problems and corresponding requirements. The processing steps corresponding to processing function 1 are defined as: 1. analyze the problem image in combination with the kill, confirm the existing problems; 2. plan the problem solution, refer to the knowledge base, and make relevant parameter changes. The output of processing function 1 is: 1. solution; 2. summary information of image problems.

[0132] The input of processing function 2 is: 1. the output of processing function 1: solution and summary information of image problems; 2. iteration indicators and 3. user requirements. The processing steps corresponding to processing function 2 are defined as: 1. iteration indicator analysis and understanding; 2. user requirement analysis and understanding; 3. processing function 1 solution understanding; 4. further analysis of matching the solution to the knowledge base and image problem summary, and summary of related optimization; 5. confirm the degree of completion, reference knowledge and unhandled cases. The output of processing function 2 is: 1. solution; 2. unhandled problems.

[0133] The execution result corresponding to processing function 2 of the solution generation unit is determined as the execution result corresponding to the solution generation unit. The solution and unhandled problems are the final goals of the solution generation unit.

[0134] The optimization unit has two processing functions, and the input of processing function 1 is: 1. the solution output by processing function 2 of the solution generation unit and unhandled problems and 2. iteration indicators. The processing steps corresponding to processing function 1 are defined as: 1. solution disassembly; 2. function call solution test and effect feedback; 3. monitor whether the effect feedback meets the standard, and automatically call analysis and optimization again if it does not meet the standard; 4. change details and supplement updates. The output of processing function 1 is: 1. optimization scheme and application call parameters; 2. update problems and optimization report, modify details.

[0135] The input of processing function 2 is: feedback information of unhandled problems; 2. iteration indicators and 3. user feedback. The processing steps corresponding to processing function 2 are defined as: 1. disassemble the feedback information of unhandled problems; 2. function call call software and monitor feedback; 3. further analyze and think according to user feedback, and re-learn and think deep knowledge base if necessary. The output of processing function 2 is: problem alarm.

[0136] The summary result of the execution results corresponding to processing functions 1 and 2 of the optimization unit is determined as the execution result corresponding to the solution generation unit. Whether the problem is solved and the warning of unhandled problems are the final goals of the optimization unit.

[0137] In the training optimization scenario, one input of the training optimization module is the problem image, which is the on-site feedback data. Only after the problem image is fed back on-site, the on-site analysis module and the training optimization module can be executed in series. The user needs to feed back the problem image after the on-site analysis model is executed. The training optimization module iteratively optimizes according to the returned problem. It should be noted that the user does not generate problem data for each execution result of the on-site analysis model.

[0138] It can be seen that by configuring the processing unit of the training optimization module as a solution generation unit and an optimization unit, and configuring the corresponding input, output and processing steps for each processing unit, the problems existing in the optimization delivery process can be solved, the manual cost of operation and maintenance can be reduced, and the optimization direction can be flexibly adjusted according to the iterative target of the user in a self-adaptive complex scene.

[0139] It should be understood that although each step in the flowchart involved in each of the above embodiments is displayed in sequence according to the arrow, these steps are not necessarily executed in the order indicated by the arrow. Unless otherwise stated herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, at least part of the steps in the flowchart involved in each of the above embodiments can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily sequential, but can be alternately executed with at least part of other steps or steps or stages in other steps.

[0140] Based on the same inventive concept, the embodiments of the present application also provide a processing device for delivering products. The implementation scheme for solving problems provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more processing device embodiments for delivering products provided below can refer to the limitations of the processing method for delivering products described above, which will not be repeated here.

[0141] As shown in Figure 7 The embodiments of the present application provide a processing device 700 for delivering products, which includes:

[0142] A demand understanding module 701 is configured to determine a target processing module corresponding to a delivery scene of the delivery product according to the obtained delivery processing information of the delivery product. The target processing module corresponding to the delivery scene includes at least one of the following: a scheme planning module, an on-site analysis module and a training optimization module.

[0143] The data input module 702 is configured to input the obtained input data of the target processing module into the target processing module for each target processing module.

[0144] The task processing module 703 is configured to perform semantic understanding on the input data to obtain at least one processing task, distribute each processing task to a corresponding processing unit for execution to obtain an execution result corresponding to each processing task, and determine a processing result corresponding to the target processing module according to the execution result corresponding to each processing task.

[0145] In some embodiments, in the process of distributing each processing task to a corresponding processing unit for execution to obtain an execution result corresponding to each processing task, the task processing module 703 is specifically configured to:

[0146] distribute each processing task to a corresponding processing unit;

[0147] determine a processing step corresponding to a processing function according to the processing function required by the processing task, the processing unit being configured with at least one processing function, each processing function being configured with at least one processing step corresponding thereto;

[0148] process the processing task according to the processing step corresponding to the processing function by the processing unit to which the processing task is distributed, to obtain an execution result corresponding to the processing task.

[0149] In some embodiments, in the process of processing the processing task according to the processing step corresponding to the processing function by the processing unit to which the processing task is distributed to obtain an execution result corresponding to the processing task, the task processing module 703 is specifically configured to:

[0150] execute the processing step corresponding to the processing function in sequence according to an execution order of the processing step by the processing unit to which the processing task is distributed;

[0151] when the target processing step is executed, generate an execution result corresponding to the target processing step according to input data and processing reference information of the target processing step;

[0152] verify the execution result corresponding to the target processing step according to verification reference information of the target processing step to obtain a verification result;

[0153] when the verification result is a verification pass, determine the execution result corresponding to the processing task according to the execution result corresponding to the target processing step.

[0154] In some embodiments, in the process of determining the target processing module corresponding to the delivery scene to which the delivery product is located according to the obtained delivery processing information of the delivery product, the demand understanding module 701 is specifically configured to:

[0155] According to the processing module configuration information, semantic understanding is performed on the obtained delivery processing information of the delivered product to obtain requirement description content corresponding to at least one target processing module;

[0156] For each target processing module, the requirement description content of the target processing module is determined as input data of the target processing module.

[0157] In some embodiments, in terms of distributing each processing task to a corresponding processing unit for execution by the target processing module to obtain an execution result corresponding to each processing task, the task processing module 703 is specifically configured to:

[0158] When the target processing module is a scheme planning module, a project understanding task is distributed to a project understanding unit for execution; the input data of the project understanding task is processed by the project understanding unit to obtain an execution result corresponding to the project understanding task; the input data of the project understanding task includes project standard process information and text information of a product station; the execution result corresponding to the project understanding task includes current scheme text description information, historical scheme text description information, and scheme configuration text description information;

[0159] A visual understanding task is distributed to a visual understanding unit for execution; the input data of the visual understanding task is processed by the visual understanding unit to obtain an execution result corresponding to the visual understanding task; the input data of the visual understanding task includes project standard process information and image information of a product station; the execution result corresponding to the visual understanding task includes current scheme image description information, historical scheme image description information, and scheme configuration image description information;

[0160] A scheme construction task is distributed to a scheme construction unit for execution; the input data of the scheme construction task is processed by the scheme construction unit to obtain an execution result corresponding to the scheme construction task; the input data of the scheme construction task includes the execution result corresponding to the project understanding task, the execution result corresponding to the visual understanding task, and user requirement information; the execution result corresponding to the scheme construction task includes a delivery scheme.

[0161] In some embodiments, in terms of distributing each processing task to a corresponding processing unit for execution to obtain an execution result corresponding to each processing task, the task processing module 703 is specifically configured to:

[0162] When the target processing module is the on-site analysis module, the on-site analysis task is assigned to the on-site analysis unit for execution; the input data of the on-site analysis task is processed by the on-site analysis unit to obtain an execution result corresponding to the on-site analysis task; the input data of the on-site analysis task includes a current on-site detection result, a historical detection result, and corresponding user demand information, a current user demand, and user interaction feedback information; the execution result corresponding to the on-site analysis task includes an application corresponding to the on-site analysis task and a calling result.

[0163] In some embodiments, in terms of assigning each processing task to a corresponding processing unit for execution to obtain an execution result corresponding to each processing task, the task processing module 703 is specifically configured to:

[0164] When the target processing module is the training optimization module, the solution generation task is assigned to the solution generation unit for execution; the input data of the solution generation task is processed by the solution generation unit to obtain an execution result corresponding to the solution generation task; the input data of the solution generation task includes a problem image, historical problems, and corresponding demand and iteration indicators; the execution result corresponding to the solution generation task includes solution content and an unprocessable problem.

[0165] The optimization task is assigned to the optimization unit for execution; the input data of the optimization task is processed by the optimization unit to obtain an execution result corresponding to the optimization task; the input data of the optimization task includes the execution result corresponding to the solution generation task, user demand information, and user feedback; the execution result corresponding to the optimization task includes a problem alarm.

[0166] Each module in the above processing device for delivering products can be realized by software, hardware, and a combination thereof, in whole or in part. Each module described above can be embedded in or independent of a processor in a computer device in a hardware form, or can be stored in a memory in a computer device in a software form, so as to be called and executed by a processor to perform operations corresponding to each module.

[0167] In some embodiments, a computer device, which can be a server, is provided, and an internal structure diagram of the computer device can be as shown in Figure 8As shown in the figure. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through the system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control ability. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data related to the processing of the delivered product. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to realize the steps in the above-mentioned product delivery processing method.

[0168] In some embodiments, a computer device, which can be a terminal, is provided, and its internal structure diagram can be as shown in the figure. Figure 9 As shown in the figure. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through the system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control ability. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data related to the processing of the delivered product. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to realize the steps in the above-mentioned product delivery processing method. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen; the input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0169] Those skilled in the art can understand that, Figure 8 Or Figure 9The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0170] In some embodiments, a computer device is provided. The computer device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the steps in the above method embodiments are implemented.

[0171] In some embodiments, as Figure 10 The figure shows an internal structure diagram of a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0172] In some embodiments, a computer program product is provided. The computer program product includes a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0173] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions.

[0174] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. The non-volatile memory can include read-only memory (Read-Only Memory, ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric memory (Ferroelectric Random Access Memory, FRAM), phase change memory (Phase Change Memory, PCM), graphene memory, etc. The volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0175] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present application.

[0176] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be noted that for those skilled in the art, without departing from the concept of the present application, some modifications and improvements can be made, which are all within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for processing delivered products, characterized in that: include: Determining a target processing module corresponding to a delivery scenario of the delivered product based on the acquired delivery processing information of the delivered product; The target processing module corresponding to the delivery scenario includes at least one of the following: a program planning module, a field analysis module, and a training optimization module; For each target processing module, inputting the acquired input data of the target processing module into the target processing module; Performing semantic understanding on the input data by the target processing module to obtain at least one processing task; Allocating each of the processing tasks to a corresponding processing unit for execution through the target processing module to obtain an execution result corresponding to each of the processing tasks; The target processing module determines the processing result corresponding to the target processing module according to the execution result corresponding to each of the processing tasks.

2. The method according to claim 1, characterized in that The step of allocating each of the processing tasks to a corresponding processing unit for execution by the target processing module to obtain an execution result corresponding to each of the processing tasks includes: Allocating each of the processing tasks to a corresponding processing unit through the target processing module; The processing unit to which the processing task is assigned determines, based on the processing function required by the processing task, a processing step corresponding to the processing function; the processing unit is configured with at least one processing function, and each processing function is configured with at least one corresponding processing step; The processing unit to which the processing task is allocated processes the processing task according to the processing steps corresponding to the processing function to obtain an execution result corresponding to the processing task.

3. The method according to claim 2, characterized in that The processing unit to which the processing task is allocated processes the processing task according to the processing steps corresponding to the processing function to obtain an execution result corresponding to the processing task, including: The processing units allocated by the processing tasks execute the processing steps in sequence according to the execution order of the processing steps corresponding to the processing functions; When executing a target processing step, generating an execution result corresponding to the target processing step according to the input data and processing reference information of the target processing step; Verifying the execution result corresponding to the target processing step according to the verification reference information of the target processing step to obtain a verification result; When the verification result is passed, the execution result corresponding to the processing task is determined according to the execution result corresponding to the target processing step.

4. The method according to claim 1, wherein The step of determining, based on the acquired delivery processing information of the delivered product, a target processing module corresponding to the delivery scenario in which the delivered product is located, includes: The demand understanding module performs semantic understanding on the acquired delivery processing information of the delivered product based on the processing module configuration information to obtain the demand description content corresponding to at least one target processing module; For each target processing module, the requirement description content of the target processing module is determined as input data of the target processing module.

5. The method according to claim 1, wherein The step of allocating each of the processing tasks to a corresponding processing unit for execution by the target processing module to obtain an execution result corresponding to each of the processing tasks includes: When the target processing module is the solution planning module, the project understanding task is assigned to the project understanding unit for execution through the solution planning module; the project understanding unit processes the input data of the project understanding task to obtain the execution result corresponding to the project understanding task; the input data of the project understanding task includes: project standard process information and text information of product workstations; the execution result corresponding to the project understanding task includes: current solution text description information, historical solution text description information and solution configuration text description information; The solution planning module assigns a visual understanding task to a visual understanding unit for execution; the visual understanding unit processes the input data of the visual understanding task to obtain an execution result corresponding to the visual understanding task; the input data of the visual understanding task includes: project standard process information and product workstation image information; the execution result corresponding to the visual understanding task includes: current solution image description information, historical solution image description information, and solution configuration image description information; The solution planning module assigns the solution construction task to the solution construction unit for execution; the solution construction unit processes the input data of the solution construction task to obtain the execution result corresponding to the solution construction task; the input data of the solution construction task includes: the execution result corresponding to the project understanding task, the execution result corresponding to the visual understanding task and user demand information; the execution result corresponding to the solution construction task includes: delivery plan.

6. The method according to claim 1, characterized in that The step of allocating each of the processing tasks to a corresponding processing unit for execution by the target processing module to obtain an execution result corresponding to each of the processing tasks includes: When the target processing module is the field analysis module, the field analysis task is assigned to the field analysis unit for execution through the field analysis module; the input data of the field analysis task is processed by the field analysis unit to obtain the execution result corresponding to the field analysis task; the input data of the field analysis task includes: current field detection results, historical detection results and corresponding user demand information, current user demand and user interaction feedback information; the execution result corresponding to the field analysis task includes: calling the application corresponding to the field analysis task and the calling result.

7. The method according to claim 1, characterized in that The step of allocating each of the processing tasks to a corresponding processing unit for execution by the target processing module to obtain an execution result corresponding to each of the processing tasks includes: When the target processing module is the training optimization module, the training optimization module assigns the solution generation task to the solution generation unit for execution; the solution generation unit processes the input data of the solution generation task to obtain the execution result corresponding to the solution generation task; the input data of the solution generation task includes: problem images, historical problems, and corresponding requirements and iteration indicators; the execution result corresponding to the solution generation task includes solution content and unprocessable problems; The optimization task is assigned to the optimization unit for execution through the training optimization module; the input data of the optimization task is processed by the optimization unit to obtain the execution result corresponding to the optimization task; the input data of the optimization task includes: the execution result corresponding to the solution generation task, user demand information and user feedback; the execution result corresponding to the optimization task includes problem alarm.

8. A processing device for delivering products, characterized in that include: A demand understanding module is used to determine a target processing module corresponding to the delivery scenario of the delivered product based on the acquired delivery processing information of the delivered product; The target processing module corresponding to the delivery scenario includes at least one of the following: a program planning module, a field analysis module, and a training optimization module; a data input module, configured to input the acquired input data of the target processing module into the target processing module; The task processing module is used to perform semantic understanding on the input data to obtain at least one processing task; assign each processing task to a corresponding processing unit for execution to obtain an execution result corresponding to each processing task; and determine the processing result corresponding to the target processing module based on the execution result corresponding to each processing task.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images