Task allocation model construction method and system for heterogeneous agents

Abstract

The invention provides a task allocation model construction method and system for heterogeneous agents, and the method comprises the steps: determining a target heterogeneous agent in real time, and carrying out the real-time matching of a target performance parameter corresponding to the target heterogeneous agent in a preset database; creating a corresponding terminology lexicon in real time according to the target performance parameters based on a preset rule, and converting the terminology lexicon into a plurality of corresponding distributed vectors in real time through a preset continuous bag-of-words model; a task allocation model for a target heterogeneous agent is trained in real time according to a preset bidirectional short-term memory network and a plurality of distributed vectors, wherein each distributed vector has uniqueness. According to the method, the task allocation of the heterogeneous agents can be objectively and accurately completed through the task allocation model, and the working efficiency is correspondingly and greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence, and particularly relates to a method and system for constructing a task allocation model for heterogeneous agents. Background Art

[0002] With the progress of technology and the rapid development of productivity, people have also made remarkable progress in the field of artificial intelligence technology, and intelligent agents such as drones and industrial robots have been developed accordingly, correspondingly improving work efficiency.

[0003] Among them, in the existing intelligent agent technology field, due to the large variety of performance parameter types among existing heterogeneous agents and the large number of performance parameters of a single intelligent agent, therefore, before existing heterogeneous agents execute tasks, targeted task allocation processing needs to be carried out.

[0004] Furthermore, in the process of task allocation for heterogeneous agents in the prior art, most are based on the models of heterogeneous agents for corresponding task allocation. However, this method cannot accurately identify the specific performance parameters of heterogeneous agents, resulting in the phenomenon that the allocated tasks do not match the actual heterogeneous agents, correspondingly reducing work efficiency. Summary of the Invention

[0005] Based on this, the purpose of the present invention is to provide a method and system for constructing a task allocation model for heterogeneous agents to solve the problem that in the process of task allocation in the prior art, the allocated tasks do not match the actual heterogeneous agents, correspondingly reducing work efficiency.

[0006] The first aspect of the embodiment of the present invention proposes: A method for constructing a task allocation model for heterogeneous agents, wherein the method includes: Determine a target heterogeneous agent in real time, and match in real time in a preset database the target performance parameters corresponding to the target heterogeneous agent; Based on preset rules, create a corresponding professional term thesaurus in real time according to the target performance parameters, and convert the professional term thesaurus into corresponding distributed vectors in real time through a preset continuous bag-of-words model; Train a task allocation model for the target heterogeneous agent in real time according to a preset bidirectional short-term memory network and a plurality of the distributed vectors, and each of the distributed vectors is unique.

[0007] The beneficial effects of the present invention are as follows: By determining the required target heterogeneous agents in real time, the corresponding processing objects can be identified. Based on this, the corresponding target performance parameters can be re-matched. Specifically, for the convenience of subsequent model construction, a corresponding professional term library can be created in real time according to the preset rules based on the current target performance parameters, and the corresponding distributed vectors can be generated synchronously. Based on this, the final model training can be carried out, and the required task allocation model can be obtained accordingly. Thus, the task allocation of each heterogeneous agent can be objectively and accurately completed through this task allocation model, and the phenomenon of mismatch can be avoided, which significantly improves the work efficiency.

[0008] Further, the step of creating a corresponding professional term library according to the target performance parameters based on preset rules includes: When the target performance parameters are obtained in real time, a comprehensive scan is performed on the target performance parameters to detect in real time several parameter types included in the target performance parameters; A corresponding target identifier is added to each parameter type, and a parameter subset corresponding to each parameter type is matched in the target performance parameters according to each target identifier; A corresponding professional term library is created in real time according to each parameter type and its corresponding parameter subset.

[0009] Further, the step of creating a corresponding professional term library in real time according to each parameter type and its corresponding parameter subset includes: When each parameter type and its corresponding parameter subset are obtained in real time, several parameter categories adapted to the parameter type are extracted in real time from the parameter subset; The parameter keywords corresponding to each parameter category are detected in real time, and the professional term library is created according to the parameter type and the corresponding parameter keywords.

[0010] Further, the step of creating the professional term library according to the parameter type and the corresponding parameter keywords includes: When each parameter type and its corresponding several parameter keywords are detected in real time, an integration process is performed on each parameter type to create a corresponding parameter string in real time; A mapping relationship between each parameter type and its corresponding several parameter keywords is constructed in real time, and each parameter keyword is mapped to the parameter string according to the mapping relationship to generate a corresponding parameter tree diagram in real time; Convert the parameter tree diagram into a corresponding glossary in real time, and create the glossary of technical terms in real time according to the glossary.

[0011] Further, the step of training a task allocation model for the heterogeneous agent based on the preset bidirectional short-term memory network and a plurality of the distributed vectors includes: When a plurality of the distributed vectors are obtained in real time, create a corresponding distributed array according to the plurality of the distributed vectors in real time; Input the distributed array into a preset embedding layer correspondingly, so that the preset embedding layer outputs a corresponding embedding vector array in real time; Train the task allocation model of the heterogeneous agent according to the embedding vector array and the preset bidirectional short-term memory network correspondingly.

[0012] Further, the step of training the task allocation model of the heterogeneous agent according to the embedding vector array and the preset bidirectional short-term memory network correspondingly includes: When the embedding vector array is obtained in real time, perform one iteration training on the preset bidirectional short-term memory network through the embedding vector array to output a corresponding initial allocation model in real time; Input the initial allocation model into a preset aggregation layer and a preset fully connected layer in sequence, and perform secondary iteration training to generate the task allocation model correspondingly.

[0013] Further, the step of inputting the initial allocation model into a preset aggregation layer and a preset fully connected layer in sequence, and performing secondary iteration training to generate the task allocation model correspondingly includes: When the initial allocation model is obtained in real time, perform aggregation processing on the initial allocation model through the preset aggregation layer, and perform global pooling processing on the initial allocation model through the preset fully connected layer; Output the task allocation model through a preset output layer correspondingly.

[0014] The second aspect of the embodiment of the present invention proposes: A task allocation model construction system for heterogeneous agents, wherein the system includes: A matching module, configured to determine a target heterogeneous agent in real time, and match a target performance parameter corresponding to the target heterogeneous agent in a preset database in real time; A creation module, configured to create a corresponding glossary of technical terms in real time according to the target performance parameter based on a preset rule, and convert the glossary of technical terms into a corresponding plurality of distributed vectors in real time through a preset continuous bag-of-words model; A training module, configured to train a task allocation model for the target heterogeneous agent in real time according to a preset bidirectional short-term memory network and a plurality of the distributed vectors, and each of the distributed vectors is unique.

[0015] Further, the creation module is specifically configured to: When the target performance parameter is obtained in real time, perform a full scan on the target performance parameter to detect in real time a plurality of parameter types included in the target performance parameter; Add corresponding target identifiers to each of the parameter types, and match parameter subsets corresponding to each of the parameter types in the target performance parameter according to each of the target identifiers; Create corresponding professional term dictionaries in real time according to each of the parameter types and the parameter subsets corresponding thereto.

[0016] Further, the creation module is specifically configured to: When each of the parameter types and the parameter subsets corresponding thereto are obtained in real time, extract a plurality of parameter categories adapted to the parameter type in the parameter subsets in real time; Detect in real time parameter keywords corresponding to each of the parameter categories, and create the professional term dictionary according to the parameter type and the corresponding parameter keywords.

[0017] Further, the creation module is specifically configured to: When each of the parameter types and a plurality of parameter keywords corresponding thereto are detected in real time, perform an integration process on each of the parameter types to create corresponding parameter strings in real time; Construct a mapping relationship between each of the parameter types and a plurality of parameter keywords corresponding thereto in real time, and map each of the parameter keywords to the parameter string according to the mapping relationship to generate a corresponding parameter tree diagram in real time; Convert the parameter tree diagram into a corresponding professional term comparison table in real time, and create the professional term dictionary in real time according to the professional term comparison table.

[0018] Further, the training module is specifically configured to: When a plurality of the distributed vectors are obtained in real time, create a corresponding distributed array in real time according to the plurality of the distributed vectors; Input the distributed array into a preset embedding layer correspondingly, so that the preset embedding layer outputs a corresponding embedding vector array in real time; Train the task allocation model of the heterogeneous agent according to the embedding vector array and the preset bidirectional short-term memory network.

[0019] Further, the training module is specifically configured to: When the embedding vector array is obtained in real time, perform an iterative training on the preset bidirectional short-term memory network through the embedding vector array to output a corresponding initial allocation model in real time; Input the initial allocation model into a preset aggregation layer and a preset fully connected layer in sequence, and perform a secondary iterative training to correspondingly generate the task allocation model.

[0020] Further, the training module is specifically configured to: When the initial allocation model is obtained in real time, perform an aggregation process on the initial allocation model through the preset aggregation layer, and perform a global pooling process on the initial allocation model through the preset fully connected layer; Output the task allocation model through a preset output layer.

[0021] The third aspect of the embodiments of the present invention proposes: A computer, including a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the computer program, the method for constructing a task allocation model for heterogeneous agents as described above is implemented.

[0022] The fourth aspect of the embodiments of the present invention proposes: A readable storage medium, on which a computer program is stored, wherein when the program is executed by a processor, the method for constructing a task allocation model for heterogeneous agents as described above is implemented.

[0023] The additional aspects and advantages of the present invention will be partially given in the following description, partially become apparent from the following description, or be understood through the practice of the present invention. Description of the Drawings

[0024] Figure 1 It is a flowchart of the method for constructing a task allocation model for heterogeneous agents provided by the first embodiment of the present invention; Figure 2 It is a structural block diagram of the system for constructing a task allocation model for heterogeneous agents provided by the third embodiment of the present invention.

[0025] The following specific embodiments will further illustrate the present invention in conjunction with the above-mentioned drawings. Specific Embodiments

[0026] To facilitate the understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0027] It should be noted that when an element is referred to as being "fixedly provided on" another element, it can be directly on the other element or there can also be an intermediate element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be an intermediate element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are only for illustrative purposes.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0029] Please refer to Figure 1 , which shows a method for constructing a task allocation model for heterogeneous agents provided by the first embodiment of the present invention. The method for constructing a task allocation model for heterogeneous agents provided by this embodiment can objectively and accurately construct the required task allocation model, and can respectively allocate appropriate tasks to each heterogeneous agent through this task allocation model, correspondingly greatly improving the work efficiency.

[0030] Specifically, this embodiment provides: A method for constructing a task allocation model for heterogeneous agents, specifically including the following steps: Step S10, real-time determine the target heterogeneous agent, and real-time match the target performance parameters corresponding to the target heterogeneous agent in the preset database; Step S20, based on preset rules, real-time create a corresponding professional term thesaurus according to the target performance parameters, and real-time convert the professional term thesaurus into corresponding several distributed vectors through a preset continuous bag-of-words model; Step S30, real-time train a task allocation model for the target heterogeneous agent according to a preset bidirectional short-term memory network and several of the distributed vectors, and each of the distributed vectors is unique.

[0031] Specifically, in this embodiment, it should be noted first that in order to objectively and effectively train the required task allocation model, accurate training data needs to be obtained in real time. Among them, it should be noted that the performance parameters of existing heterogeneous agents can directly determine the target tasks they can handle, that is, the tasks that existing heterogeneous agents can handle are related to their performance parameters. Based on this, in order to objectively and accurately complete the task allocation in the actual application process, the corresponding processing object needs to be determined first, that is, the present invention will first determine a target heterogeneous agent for processing. Specifically, the target heterogeneous agent can be devices such as unmanned aerial vehicles and industrial robots. Based on this, the present invention can first match the target performance parameters corresponding to the current target heterogeneous agent in the existing database that has been created, that is, the data processing ability of the current target heterogeneous agent, for subsequent processing.

[0032] Further, after obtaining the required target performance parameters in real time through the above steps, the present invention can be based on the current target performance parameters. At the same time, according to the pre-set rules, a corresponding professional term thesaurus is immediately generated in real time according to the current target performance parameters. Based on this, in order to obtain the data for subsequent training, the current professional term thesaurus is further processed in real time by the existing continuous bag-of-words model, and several distributed vectors can be generated again. Specifically, the several distributed vectors can be directly used as the data for subsequent training. Thus, the present invention can finally train the task allocation model for the above-mentioned heterogeneous agents in real time according to the pre-set bidirectional short-term memory network and the current several distributed vectors. And in the actual application process, the task allocation model can automatically complete the task allocation of each heterogeneous agent, thereby avoiding the phenomenon that the allocated tasks do not match the processing capabilities of the heterogeneous agents, and correspondingly greatly improving the work efficiency.

[0033] Second Embodiment Further, the step of creating a corresponding professional term thesaurus according to the target performance parameters based on the preset rules includes: When the target performance parameters are obtained in real time, a full scan is performed on the target performance parameters to detect several parameter types contained in the target performance parameters in real time; A corresponding target identifier is added to each parameter type, and a parameter subset corresponding to each parameter type is matched in the target performance parameters according to each target identifier; A corresponding professional term thesaurus is created in real time according to each parameter type and its corresponding parameter subset.

[0034] Further, the step of creating a corresponding professional term thesaurus in real time according to each parameter type and its corresponding parameter subset includes: When each parameter type and its corresponding parameter subset are obtained in real time, several parameter categories adapted to the parameter type are extracted in real time from the parameter subset; Parameter keywords corresponding to each parameter category are detected in real time, and the professional term thesaurus is created correspondingly according to the parameter type and the corresponding parameter keywords.

[0035] Further, the step of creating the professional term thesaurus correspondingly according to the parameter type and the corresponding parameter keywords includes: When each parameter type and several corresponding parameter keywords are detected in real time, each parameter type is integrally processed to create a corresponding parameter string in real time; A mapping relationship between each parameter type and several corresponding parameter keywords is constructed in real time, and each parameter keyword is mapped to the parameter string according to the mapping relationship to generate a corresponding parameter tree diagram in real time; The parameter tree diagram is converted into a corresponding professional term comparison table in real time, and the professional term thesaurus is created in real time according to the professional term comparison table.

[0036] Further, the step of training a task allocation model for the heterogeneous agent in real time according to the preset bidirectional short-term memory network and several distributed vectors includes: When several distributed vectors are obtained in real time, a corresponding distributed array is created in real time according to the several distributed vectors; The distributed array is correspondingly input into a preset embedding layer so that the preset embedding layer outputs a corresponding embedding vector array in real time; The task allocation model of the heterogeneous agent is trained correspondingly according to the embedding vector array and the preset bidirectional short-term memory network.

[0037] Further, the step of training the task allocation model of the heterogeneous agent correspondingly according to the embedding vector array and the preset bidirectional short-term memory network includes: When the embedding vector array is obtained in real time, the preset bidirectional short-term memory network is iteratively trained once through the embedding vector array to output a corresponding initial allocation model in real time; The initial allocation model is sequentially input into a preset aggregation layer and a preset fully connected layer, and is iteratively trained a second time to correspondingly generate the task allocation model.

[0038] Further, the step of inputting the initial allocation model into a preset aggregation layer and a preset fully-connected layer in sequence and performing secondary iterative training to correspondingly generate the task allocation model includes: When the initial allocation model is obtained in real time, perform aggregation processing on the initial allocation model through the preset aggregation layer, and perform global pooling processing on the initial allocation model through the preset fully-connected layer; Correspondingly output the task allocation model through a preset output layer.

[0039] In addition, in this embodiment, it should also be noted that after obtaining the required target performance parameters in real time through the above steps, it is necessary to parse the current target performance parameters in real time to generate a professional term library for subsequent improving the training efficiency. Specifically, in order to comprehensively obtain all the information contained in the current target performance parameters, the present invention will immediately perform a full scan of the current target performance parameters. At the same time, several parameter types contained therein can be detected in real time within the current target performance parameters. Specifically, for the sake of understanding, for example, these several parameter types can be mechanical parameters and software parameters, etc. Based on this, parameter subsets corresponding to each current parameter type can be matched again in real time within the above target performance parameters. Specifically, for the sake of implementation, the present invention will parse the current parameter subsets again and can extract several parameter categories adapted to the current parameter type from within the parameter subsets. Among them, for the sake of understanding, these several parameter categories can be mechanical parameters and dynamic parameters, etc. Based on this, the present invention will take the current several parameter categories as the processing objects again and can detect the parameter keywords corresponding to each current parameter category in real time. The parameter keywords can correspondingly represent the information contained in the current parameter category. Based on this, for the sake of integrating the current various data, the present invention will first perform integration processing on each current parameter type and can create corresponding parameter strings in real time according to the character information contained within each current parameter type. At the same time, for the sake of subsequent integration, the present invention will also construct a mapping relationship between each current parameter type and the several parameter keywords corresponding thereto in real time. Based on this, the present invention will directly map each current parameter keyword to the above parameter strings one by one according to this mapping relationship, so as to be able to form a corresponding parameter tree diagram in real time. Specifically, this parameter tree diagram can clearly reflect the mapping relationship between the parameter types and each parameter category, so that the distribution of professional terms corresponding to each parameter category can be clearly obtained. Furthermore, the current parameter tree diagram can be converted into a corresponding professional term comparison table in real time, and a corresponding professional term library can be created in real time according to this professional term comparison table for subsequent processing.

[0040] Furthermore, after obtaining a number of required distributed vectors in real time through the above steps, at this time, the current number of distributed vectors can be arrayed, and a corresponding distributed array can be formed. Specifically, the present invention will input the current distributed array into a pre-set embedding layer in real time, and can convert the current distributed array into an embedding vector array for subsequent training through the pre-set embedding layer. Based on this, in order to quickly and effectively complete the subsequent model training, the present invention will immediately perform an iterative training on the current pre-set bidirectional short-term memory network through the current embedding vector array, and can correspondingly train the required initial allocation model. Then, the current initial allocation model will be input into the pre-set aggregation layer and the fully connected layer in sequence again. During this process, the current aggregation layer will perform an aggregation process on the current initial allocation model. Correspondingly, the current fully connected layer will also perform a global pooling process on the current initial allocation model for corresponding secondary iterative training, so as to finally train the required task allocation model and complete the task allocation through this task allocation model, improving work efficiency.

[0041] Please refer to Figure 2 , the third embodiment of the present invention provides: A task allocation model construction system for heterogeneous agents, wherein the system includes: A matching module, configured to determine a target heterogeneous agent in real time and match the target performance parameters corresponding to the target heterogeneous agent in a preset database in real time; A creation module, configured to create a corresponding professional term library in real time based on the target performance parameters according to preset rules, and convert the professional term library into a corresponding number of distributed vectors in real time through a preset continuous bag-of-words model; A training module, configured to train a task allocation model for the target heterogeneous agent in real time according to a preset bidirectional short-term memory network and a number of the distributed vectors, and each of the distributed vectors is unique.

[0042] Furthermore, the creation module is specifically configured to: When the target performance parameters are obtained in real time, perform a full scan on the target performance parameters to detect a number of parameter types included in the target performance parameters in real time; Add corresponding target identifiers to each of the parameter types, and match a parameter subset corresponding to each of the parameter types in the target performance parameters according to each of the target identifiers; Create a corresponding professional term library according to each of the parameter types and its corresponding parameter subset in real time.

[0043] Furthermore, the creation module is specifically configured to: When each of the parameter types and its corresponding parameter subset are obtained in real time, several parameter categories adapted to the parameter type are extracted in real time from the parameter subset; Parameter keywords corresponding to each of the parameter categories are detected in real time, and the professional term thesaurus is created correspondingly according to the parameter type and the corresponding parameter keywords.

[0044] Further, the creation module is specifically configured to: When each of the parameter types and several corresponding parameter keywords are detected in real time, each of the parameter types is integrally processed to create a corresponding parameter string in real time; A mapping relationship between each of the parameter types and several corresponding parameter keywords is constructed in real time, and each of the parameter keywords is mapped to the parameter string according to the mapping relationship to generate a corresponding parameter tree diagram in real time; The parameter tree diagram is converted into a corresponding professional term comparison table in real time, and the professional term thesaurus is created in real time according to the professional term comparison table.

[0045] Further, the training module is specifically configured to: When several of the distributed vectors are obtained in real time, a corresponding distributed array is created in real time according to the several distributed vectors; The distributed array is correspondingly input into a preset embedding layer so that the preset embedding layer outputs a corresponding embedding vector array in real time; The task allocation model of the heterogeneous agent is trained according to the embedding vector array and the preset bidirectional short-term memory network.

[0046] Further, the training module is specifically configured to: When the embedding vector array is obtained in real time, the preset bidirectional short-term memory network is iteratively trained once through the embedding vector array to output a corresponding initial allocation model in real time; The initial allocation model is sequentially input into a preset aggregation layer and a preset fully connected layer, and secondary iterative training is performed to correspondingly generate the task allocation model.

[0047] Further, the training module is specifically configured to: When the initial allocation model is obtained in real time, the initial allocation model is aggregated through the preset aggregation layer, and global pooling processing is performed on the initial allocation model through the preset fully connected layer; The task allocation model is output correspondingly through a preset output layer.

[0048] The fourth embodiment of the present invention provides a computer, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method for constructing a task allocation model for heterogeneous agents as described above is implemented.

[0049] The fifth embodiment of the present invention provides a readable storage medium, on which a computer program is stored. When the program is executed by a processor, the method for constructing a task allocation model for heterogeneous agents as described above is implemented.

[0050] In summary, the method and system for constructing a task allocation model for heterogeneous agents provided in the above embodiments of the present invention can objectively and accurately train the required task allocation model, and can complete the task allocation of each heterogeneous agent through the task allocation model, correspondingly improving the work efficiency.

[0051] It should be noted that the above-mentioned each module can be a functional module or a program module, and can be implemented either by software or by hardware. For the modules implemented by hardware, the above-mentioned each module can be located in the same processor; or the above-mentioned each module can also be located in different processors in any combined form.

[0052] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in conjunction with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0053] More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other appropriate processing as necessary, and then stored in a computer memory.

[0054] It should be understood that each part of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following technologies well known in the art can be used: discrete logic circuits with logic gate circuits for implementing logical functions on data signals, application specific integrated circuits with appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0055] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0056] The above-described embodiments merely represent several implementation manners of the present invention, and their descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the appended claims.

Claims

1. A method for constructing a task allocation model for heterogeneous agents, characterized in that: The method comprises: Determine the target heterogeneous intelligent agent in real time, and match the target performance parameters corresponding to the target heterogeneous intelligent agent in real time in a preset database; Based on preset rules, a corresponding professional terminology word library is created in real time according to the target performance parameters, and the professional terminology word library is converted into corresponding distributed vectors in real time through a preset continuous bag-of-words model; A task allocation model for the target heterogeneous agent is trained in real time based on a preset bidirectional short-term memory network and a plurality of the distributed vectors, and each of the distributed vectors is unique.

2. The method for constructing a task allocation model for heterogeneous agents according to claim 1, characterized in that: The step of creating a corresponding professional terminology vocabulary in real time according to the target performance parameters based on preset rules includes: When the target performance parameter is acquired in real time, the target performance parameter is fully scanned to detect in real time several parameter types corresponding to the target performance parameter; Adding a corresponding target identifier to each parameter type, and matching a parameter subset corresponding to each parameter type in the target performance parameters according to each target identifier; A corresponding professional terminology word library is created in real time according to each parameter type and its corresponding parameter subset.

3. The method for constructing a task allocation model for heterogeneous agents according to claim 2, characterized in that: The step of creating a corresponding professional terminology vocabulary in real time according to each parameter type and its corresponding parameter subset comprises: When each parameter type and its corresponding parameter subset are acquired in real time, a plurality of parameter types matching the parameter type are extracted in real time from the parameter subset; The parameter keywords corresponding to each parameter type are detected in real time, and the professional terminology dictionary is created according to the parameter type and the corresponding parameter keywords.

4. The method for constructing a task allocation model for heterogeneous agents according to claim 3, characterized in that: The step of creating the professional terminology dictionary according to the parameter type and the corresponding parameter keyword includes: When each parameter type and a number of parameter keywords corresponding thereto are detected in real time, each parameter type is integrated and processed to create a corresponding parameter string in real time; Constructing in real time a mapping relationship between each parameter type and a number of corresponding parameter keywords, and mapping each parameter keyword to the parameter string according to the mapping relationship, so as to generate a corresponding parameter tree diagram in real time; The parameter tree diagram is converted into a corresponding professional term comparison table in real time, and the professional term vocabulary is created in real time according to the professional term comparison table.

5. The method for constructing a task allocation model for heterogeneous agents according to claim 1, characterized in that: The step of training the task allocation model for the heterogeneous agents in real time according to the preset bidirectional short-term memory network and the plurality of distributed vectors comprises: When a number of the distributed vectors are acquired in real time, a corresponding distributed array is created in real time according to the number of the distributed vectors; Inputting the distributed array into a preset embedding layer accordingly, so that the preset embedding layer outputs a corresponding embedding vector array in real time; A task allocation model of the heterogeneous agent is trained according to the embedding vector array and the preset bidirectional short-term memory network.

6. The method for constructing a task allocation model for heterogeneous agents according to claim 5, characterized in that: The step of training the task allocation model of the heterogeneous agent according to the embedding vector array and the preset bidirectional short-term memory network comprises: When the embedding vector array is acquired in real time, the preset bidirectional short-term memory network is iteratively trained once through the embedding vector array to output the corresponding initial allocation model in real time; The initial allocation model is sequentially input into a preset aggregation layer and a preset fully connected layer, and a second iterative training is performed to generate the task allocation model accordingly.

7. The method for constructing a task allocation model for heterogeneous agents according to claim 6, characterized in that: The step of inputting the initial allocation model into a preset aggregation layer and a preset fully connected layer in sequence and performing secondary iterative training to generate the task allocation model accordingly includes: When the initial allocation model is acquired in real time, the initial allocation model is aggregated by the preset aggregation layer, and the initial allocation model is globally pooled by the preset fully connected layer; The task allocation model is outputted correspondingly through a preset output layer.

8. A task allocation model construction system for heterogeneous agents, characterized in that: The system comprises: A matching module is used to determine the target heterogeneous intelligent agent in real time and match the target performance parameters corresponding to the target heterogeneous intelligent agent in real time in a preset database; A creation module, used to create a corresponding professional terminology word library in real time according to the target performance parameters based on preset rules, and convert the professional terminology word library into corresponding distributed vectors in real time through a preset continuous bag of words model; A training module is used to train a task allocation model for the target heterogeneous agent in real time based on a preset bidirectional short-term memory network and a plurality of the distributed vectors, each of which is unique.

9. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements the method for constructing a task allocation model for heterogeneous agents as described in any one of claims 1 to 7.

10. A readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, it implements the method for constructing a task allocation model for heterogeneous agents as described in any one of claims 1 to 7.

Images