PID (Proportion Integration Differentiation) control system setting method and device

Abstract

The invention relates to the field of air conditioners, and provides a PID control system setting method and device, and the method comprises the steps: obtaining setting demand data; word vector extraction is carried out on the setting demand data to obtain word vector key value pairs; performing similarity matching on the word vector key value pair and each word vector in a word vector library, and determining a PID setting parameter based on a similarity matching result; wherein the word vector library is obtained by performing word vector extraction on the basis of the constructed knowledge graph library, and the knowledge graph library is constructed on the basis of the setting item entity, the to-be-set equipment entity and the relationship between the setting item entity and the to-be-set equipment entity. According to the method, the defect that the PID setting effect is poor due to the fact that expert knowledge and experience are manually converted is overcome, the PID setting parameters are rapidly and accurately positioned by combining the knowledge graph database, and the process of determining the PID setting parameters is accelerated.

Description

Technical Field

[0001] The present invention relates to the technical field of air conditioners, and in particular, to a method and device for tuning a PID control system. Background Art

[0002] The tuning of a Proportional Integral Derivative (PID) control system has always been an important topic in control theory. In theoretical research, there are many methods for PID tuning, including theoretical modeling methods, machine learning, reinforcement learning, deep learning methods, etc. The essence of these methods is to model the controlled object through a white-box mechanism model, a black-box model, or a gray-box model, so as to achieve PID tuning. If these theoretical methods are to achieve good results, the debugging personnel need to have a strong theoretical background and cross-disciplinary knowledge. Therefore, in practical applications, especially in HVAC systems, such as high-efficiency HVAC machine rooms, etc., the expert experience method is often used to tune the PID controller.

[0003] However, due to the strong time delay of the HVAC system, the mutual influence between various devices, the large modeling difficulty between control parameters and observation parameters, the long debugging cycle of the HVAC system, the debugging effect depending on the conversion of expert knowledge by the debugging personnel, and the huge difference in performance in different projects after the conversion of expert knowledge, and the control failure caused by the easy deviation of the device sensor in the later stage, etc., many factors lead to poor tuning effect of the expert knowledge and experience debugging in the equipment field with strong time delay and mutual influence between various devices, consuming a large amount of manpower, material resources and financial resources, and unable to meet the control requirements. Summary of the Invention

[0004] The present invention provides a method and device for tuning a PID control system, which is used to solve the defect of poor PID tuning effect caused by manual conversion of expert knowledge and experience in the prior art, and quickly and accurately locate the PID tuning parameters by combining with a knowledge graph library, so as to speed up the process of determining the PID tuning parameters.

[0005] The present invention provides a method for tuning a PID control system, including: obtaining tuning requirement data; extracting word vectors from the tuning requirement data to obtain word vector key-value pairs; performing similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determining the PID tuning parameters based on the similarity matching result; wherein, the word vector library is obtained by extracting word vectors based on the constructed knowledge graph library in advance, and the knowledge graph library is constructed based on the tuning project entity, the device to be tuned entity, and the relationship between the tuning project entity and the device to be tuned entity.

[0006] It should be noted that by extracting word vectors from the obtained setting requirement data, the data in text form is transformed into vector representation, so as to better capture the semantic relationships and potential features between words, more accurately mine the key information contained in the setting requirement data, avoid the deviation of demand understanding caused by the surface differences of the text, and further perform similarity matching between the extracted word vector key-value pairs and each word vector in the word vector library. According to the similarity degree at the semantic level, the relevant content in the knowledge graph that best matches the setting requirements is found, and then the appropriate PID tuning parameters are accurately determined, making the process of determining PID tuning parameters more scientific and accurate. In addition, a knowledge graph library is constructed based on the setting project entity, the device to be tuned entity, and the relationship between the two, so as to systematically integrate the scattered and fragmented setting knowledge, and further extract a word vector library based on the knowledge graph library, enabling the knowledge in the knowledge graph to be quickly retrieved and matched in vector form, facilitating the efficient mining of relevant historical experience and knowledge from the word vector library, quickly locating useful information, improving the efficiency of knowledge utilization, and helping to speed up the entire process of determining PID tuning parameters.

[0007] According to the PID control system tuning method provided by the present invention, similarity matching is performed between the word vector key-value pairs and each word vector in the word vector library, and based on the similarity matching result, PID tuning parameters are determined, including: performing similarity matching between the word vector key-value pairs and each word vector in the word vector library to obtain a similarity matching result, where the similarity matching result includes the similarity between the word vector key-value pairs and each word vector in the word vector library; sorting each word vector in the word vector library according to the similarity, and selecting a preset number of word vectors based on the sorting result to obtain the PID tuning parameters.

[0008] It should be noted that by performing similarity matching between the word vector key-value pairs and each word vector in the word vector library, the semantic information contained in the word vectors is fully utilized to find the word vectors in the word vector library that best match the setting requirements from the semantic level, avoiding missing key similar experiences or knowledge due to differences in text expressions, and sorting each word vector in the word vector library according to the similarity, so as to arrange the word vectors in an orderly manner according to the degree of relevance to the current demand, realizing the effective screening and focusing of a large amount of information, avoiding the decision-making confusion caused by information overload, and selecting a preset number of word vectors based on the sorting result to preferentially select the word vectors that best meet the requirements, avoiding blindly selecting or randomly intercepting part of the content among numerous word vectors, refining the most critical and valuable part, quickly locking the content with the strongest relevance to the current specific tuning scenario, and thus determining the PID parameters based on these selected information, which helps to improve the efficiency of decision-making and make the entire tuning parameter determination process more efficient and fast.

[0009] According to a PID control system tuning method provided by the present invention, a preset number of word vectors are selected based on a sorting result to obtain a PID tuning parameter, including: based on a preset number of 1, according to the sorting result, a word vector with the largest similarity is selected as the PID tuning parameter; based on a preset number greater than 1, according to the sorting result, a preset number of word vectors are selected in descending order of similarity and average value aggregation is performed to obtain the PID tuning parameter.

[0010] It is worth noting that when the preset number is 1, directly selecting the word vector with the largest similarity as the PID tuning parameter can achieve highly accurate focusing, avoid being interfered by other relatively weakly correlated information, and make the decision-making process concise and straightforward; when the preset number is greater than 1, a preset number of word vectors are selected from large to small according to the similarity and the average value is aggregated to obtain the PID tuning parameter. By aggregating the parameters corresponding to these word vectors, the fusion of multiple relevant information sources can be achieved, thereby comprehensively considering multiple situations, avoiding the one-sidedness and limitations that may be caused by relying on only a single information, and making the determined PID parameters more comprehensive and representative.

[0011] According to a PID control system tuning method provided by the present invention, word vectors are extracted from tuning requirement data to obtain word vector key-value pairs, including: inputting the tuning requirement data into an embedded Embedding model to obtain word vector key-value pairs output by the Embedding model; wherein the Embedding model is trained based on tuning requirement training data and word vector labels corresponding to the tuning requirement training data.

[0012] It should be noted that by inputting the tuning requirement data into the embedded Embedding model, the Embedding model can be used to mine the deep semantic relationship between words and map them to similar vector space positions, so that subsequent operations based on these word vectors can more accurately utilize semantic information, which helps to efficiently determine PID tuning parameters and other related content in the future, thereby improving the efficiency of the entire workflow.

[0013] According to the present invention, a PID control system tuning method is provided to obtain tuning demand data, including: obtaining the tuning demand sent by the demander; obtaining the demand inference result based on the tuning demand using the demand information prediction model; wherein the demand information prediction model is trained based on the demand training data and the result labels corresponding to the demand training data; and extracting keywords from the demand inference result, and combining it with a preset format to obtain the tuning demand data.

[0014] It should be noted that by obtaining the tuning requirements sent by the requester, directly collecting the original information closely related to the actual application scenario from the source, ensuring that the processed data is generated based on real requirements, and using the demand information prediction model to obtain the demand inference result, further mining the potential and unexpressed demands of the requester, improving the demand information, making the final tuning requirements more comprehensive and accurate, avoiding the situation that all demand details cannot be fully elaborated due to limitations in professional knowledge or insufficient understanding of the system, extracting keywords from the demand inference result to refine the core and key information content, and combining with the preset format to organize and obtain the tuning requirement data, realizing the standardization and normalization of the data format, avoiding difficulties in understanding and operation caused by format chaos, and ensuring the smooth progress of the entire tuning process.

[0015] According to the present invention, a PID control system tuning method is provided. According to the tuning requirements, using the demand information prediction model, the demand inference result is obtained, including: inputting the tuning requirements into the demand information prediction model to obtain the demand inference result output by the demand information prediction model; wherein, the demand information prediction model is trained based on the demand training data and the result labels corresponding to the demand training data; determining whether there are missing information parameters in the demand inference result, if there are missing information parameters, generating an information acquisition request based on the missing information parameters and returning it to the requester to re-obtain the tuning requirements, and inputting the re-obtained tuning requirements into the demand information prediction model to obtain the updated demand inference result output by the demand information prediction model, and re-determining whether there are missing information parameters in the updated demand inference result until there are no missing information parameters in the updated demand inference result.

[0016] It should be added that by determining whether there are missing information parameters in the demand inference result and generating an information acquisition request in a timely manner when missing is found and returning it to the requester, the demand information is actively improved, avoiding the situation that subsequent tuning work is deviated or cannot be carried out smoothly due to incomplete information, ensuring that the demand information finally used for tuning covers all necessary contents, and by continuously repeating the process of re-obtaining the tuning requirements, inputting them into the model again to obtain the updated demand inference result and re-checking the missing information parameters until there are no missing information parameters, so as to realize the multiple iterative improvement of the demand information, thereby ensuring the integrity of the demand information to the greatest extent.

[0017] According to a PID control system tuning method provided by the present invention, before similarity matching is performed between word vector key-value pairs and word vectors in a word vector library, the method includes: based on a preset tuning type, obtaining corresponding project information, and constructing a tuning project entity of the corresponding tuning type; based on a preset disturbance condition, disturbing the parameters of a device to be tuned to obtain a gain coefficient and an oscillation period of the device to be tuned; determining the PID three-term coefficients of the device to be tuned according to the gain coefficient and the oscillation period of the device to be tuned and the preset constants of the device to be tuned; constructing a device entity to be tuned corresponding to the device to be tuned according to the gain coefficient and the oscillation period of the device to be tuned, the PID three-term coefficients of the device to be tuned, and the device information of the device to be tuned previously obtained; constructing a knowledge graph library according to the relationship between the tuning project entities of each tuning type and each device entity to be tuned; extracting word vectors from the knowledge graph library, and obtaining the word vector library based on the extracted word vectors.

[0018] It should be noted that by obtaining the corresponding project information according to the preset tuning type and then constructing the tuning project entity, all subsequent analysis and processing can be carried out closely around the specific type of tuning project; by disturbing the parameters of the tuning equipment based on the preset disturbance conditions, the gain coefficient and oscillation period are obtained, which provides basic data for the subsequent reasonable setting of PID parameters to achieve precise speed regulation; by determining the three PID coefficients according to the gain coefficient, oscillation period and preset constants, the characteristics of the equipment itself and the established control-related parameter requirements are comprehensively considered, so that the calculated PID The coefficients are more in line with the actual control needs of the equipment, avoiding the problem of poor control effect that may be caused by setting parameters based on experience or general formulas; by constructing a knowledge graph library based on the relationship between the tuning project entities of each tuning type and the entities of each device to be tuned, the scattered and fragmented knowledge about the tuning projects and equipment is effectively integrated, which is convenient for technical personnel to fully and systematically understand the knowledge architecture of the entire tuning field, and provides a clear knowledge context and reference basis for complex tuning work, which helps to quickly locate and use relevant knowledge to solve practical problems; by extracting word vectors from the knowledge graph library to obtain a word vector library, the structured knowledge in the knowledge graph is further converted into a vector form that is convenient for computers to perform semantic analysis and similarity matching, so that in the subsequent similarity matching of word vector key-value pairs with the word vector library, the rich knowledge contained in the knowledge graph can be fully utilized to dig out the semantic associations between different entities and concepts, and find the information that best fits the current tuning needs from a more macro and comprehensive perspective, thereby improving the efficiency and accuracy of the entire information retrieval and utilization, and providing strong support for the accurate determination of PID tuning parameters.

[0019] The present invention also provides a PID control system tuning device, including: a data acquisition module for acquiring tuning requirement data; a word embedding module for extracting word vectors from the tuning requirement data to obtain word vector key-value pairs; a parameter determination module for performing similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determining PID tuning parameters based on the similarity matching result; wherein, the word vector library is obtained by previously extracting word vectors based on a constructed knowledge graph library, and the knowledge graph library is constructed based on tuning project entities, equipment to be tuned entities, and the relationships between the tuning project entities and the equipment to be tuned entities.

[0020] It should be noted that the word embedding module extracts word vectors from the tuning requirement data acquired by the data acquisition module to convert the data in text form into vector representation, so as to better capture the semantic relationships and potential features between words, more accurately dig out the key information contained in the tuning requirement data, avoid the deviation of demand understanding caused by the surface differences of the text, and further, the parameter determination module performs similarity matching between the extracted word vector key-value pairs and each word vector in the word vector library, so as to find the relevant content in the knowledge graph that best matches the tuning requirements according to the similarity degree at the semantic level, and then accurately determine the appropriate PID tuning parameters, making the process of determining PID tuning parameters more scientific and accurate. In addition, a knowledge graph library is constructed based on tuning project entities, equipment to be tuned entities, and the relationships between the two, so as to systematically integrate the scattered and fragmented tuning knowledge, and further extract a word vector library based on the knowledge graph library, enabling the knowledge in the knowledge graph to be quickly retrieved and matched in vector form, thus facilitating the efficient mining of relevant historical experience and knowledge from the word vector library, quickly locating useful information, improving the efficiency of knowledge utilization, and helping to speed up the entire process of determining PID tuning parameters.

[0021] The present invention also provides an electronic device, 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 steps of any one of the above-mentioned PID control system tuning methods are implemented.

[0022] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of any one of the above-mentioned PID control system tuning methods are implemented. Description of the Drawings

[0023] To more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0024] Figure 1 is one of the schematic flowcharts of the PID control system tuning method provided by the present invention; Figure 2 is the second schematic flowchart of the PID control system tuning method provided by the present invention; Figure 3 is the schematic flowchart of determining the PID tuning parameters provided by the present invention; Figure 4 is the schematic structural diagram of the PID control system tuning device provided by the present invention; Figure 5 is the schematic structural diagram of the electronic device provided by the present invention. Detailed implementation manners

[0025] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in the present invention belong to the scope of protection of the present invention.

[0026] Figure 1 Describes the schematic flowchart of a PID control system tuning method of the present invention. The method includes: S11, obtaining tuning requirement data; S12, extracting word vectors from the tuning requirement data to obtain word vector key-value pairs; S13, performing similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determining the PID tuning parameters based on the similarity matching result; wherein, the word vector library is obtained by extracting word vectors based on the pre-constructed knowledge graph library, and the knowledge graph library is constructed based on the tuning project entity, the device to be tuned entity, and the relationship between the tuning project entity and the device to be tuned entity.

[0027] It should be noted that the step numbers "S1N" in this specification do not represent the sequence of the PID control system tuning method. The following will specifically describe Figures 2 - 3 the PID control system tuning method of the present invention.

[0028] Step S11: Obtain the setting requirement data.

[0029] In this embodiment, with reference to Figure 2 , the obtaining of the setting requirement data includes: obtaining the setting requirements sent by the requester; according to the setting requirements, using the demand information prediction model to obtain the demand inference result; wherein, the demand information prediction model is trained based on the demand training data and the result labels corresponding to the demand training data; extracting keywords from the demand inference result and combining with the preset format to obtain the setting requirement data.

[0030] It should be noted that by obtaining the setting requirements sent by the requester, the original information closely related to the actual application scenario is directly collected from the source to ensure that the processed data is generated based on real demands. The demand inference result is obtained by using the demand information prediction model to further explore the potential and unexpressed demands of the requester, improve the demand information, make the final setting requirements more comprehensive and accurate, and avoid the situation where all demand details cannot be fully elaborated due to limitations in professional knowledge or insufficient understanding of the system. Keywords are extracted from the demand inference result to refine the core and key information content, and the setting requirement data is organized by combining with the preset format to achieve the standardization and normalization of the data format, avoid difficulties in understanding and operation caused by format chaos, and ensure the smooth progress of the entire setting process.

[0031] Specifically, the demand information prediction model includes a feature extraction layer, a demand prediction layer, an information parameter extraction layer, and an information integration layer. The setting requirements are input into the demand information prediction model to obtain the demand inference result output by the demand information prediction model, including: inputting the demand information into the feature extraction layer for feature extraction to obtain the demand features output by the feature extraction layer; inputting the demand features into the demand prediction layer for demand information prediction to obtain the prediction information output by the demand prediction layer, and the prediction information includes the project information and equipment information required for the setting requirements; inputting the demand features into the demand information extraction layer for information parameter extraction to obtain the parameter information output by the demand information extraction layer; inputting the prediction information and the parameter information into the information integration layer for information integration to obtain the demand inference result output by the information integration layer.

[0032] It should be added that the setting requirements include the setting type and the project information involved in the setting type, and the demand inference result includes the geographical location, rated cooling capacity, quantity, equipment type, setting type, rated power of the equipment, rated flow rate and head of the equipment.

[0033] It should be noted that if the obtained setting requirements contain all the information parameters required for the demand inference result, the keywords can be directly extracted from the demand inference result output by the demand information prediction model and combined with the preset format to obtain the setting requirement data converted into the standard format.

[0034] A preset format can be expressed as: Args: { "location": {"latitude": xx.xx, "longitude": yy.yy}, "cap_design": xxxx, "num": a, "device_type": b, "tuner_type": c, "norm_power": d, "norm_flow": e, "norm_cap": f } Among them, "location" represents the geographical location, "cap_design" represents the rated cooling capacity, "num" represents the quantity, "device_type" represents the device type, "tuner_type" represents the setting type, "norm_power" represents the rated power of the device, "norm_flow" represents the rated flow rate of the device, and "norm_cap" represents the head. The specific values of each information parameter can be determined based on the actual design requirements and are not further limited here.

[0035] In addition, if the obtained setting requirements do not completely contain all the information parameters required for the demand inference result, then according to the setting requirements, using the demand information prediction model, the demand inference result is obtained, including: inputting the setting requirements into the demand information prediction model to obtain the demand inference result output by the demand information prediction model; among them, the demand information prediction model is trained based on the demand training data and the result labels corresponding to the demand training data; determining whether there are missing information parameters in the demand inference result. If there are missing information parameters, an information acquisition request is generated based on the missing information parameters and returned to the demand side to re-obtain the setting requirements, and the re-obtained setting requirements are input into the demand information prediction model to obtain the updated demand inference result output by the demand information prediction model, and re-determining whether there are missing information parameters in the updated demand inference result until there are no missing information parameters in the updated demand inference result.

[0036] It should be added that by determining whether there are missing information parameters in the demand reasoning result and generating an information acquisition request in a timely manner when a missing is found and returning it to the demand side, the demand information can be actively improved, avoiding deviations or failures in the subsequent setting work due to incomplete information, ensuring that the demand information finally used for setting covers all necessary contents, and by continuously repeating the process of re-obtaining the setting demand, inputting it into the model again to obtain the updated demand reasoning result and re-checking the missing information parameters until there are no missing information parameters, so as to realize multiple iterations and improvements of the demand information, thus ensuring the integrity of the demand information to the greatest extent.

[0037] In addition, the information acquisition request can be configured based on the demand information that needs to be supplemented and improved. For example, if the demand information lacks project information, project geographical location, and rated cooling capacity, the information acquisition request can be "I need to know what the project is, where it is, and what the rated cooling capacity is", and no further limitation is made here.

[0038] It should be noted that the demand information prediction model can adopt large models such as LLM and combine the logical reasoning ability of the ReAct agent to achieve multi-round question-and-answer reasoning to ensure that all information parameters required for the demand reasoning result can be obtained.

[0039] Step S12: Extract word vectors from the setting demand data to obtain word vector key-value pairs.

[0040] In this embodiment, extracting word vectors from the setting demand data to obtain word vector key-value pairs includes: inputting the setting demand data into an embedded Embedding model to obtain the word vector key-value pairs output by the Embedding model; where the Embedding model is trained based on the setting demand training data and the corresponding word vector labels of the setting demand training data.

[0041] It should be noted that by inputting the setting demand data into the embedded Embedding model, the deep semantic relationship between words can be mined by the Embedding model and mapped to a similar vector space position, so that the subsequent operations based on these word vectors can more accurately utilize the semantic information, which helps to efficiently determine relevant contents such as PID setting parameters in the subsequent process and improve the efficiency of the entire work process.

[0042] Specifically, the Embedding model includes a vector extraction layer and a feature fusion layer. The setting requirement data is input into the embedded Embedding model to obtain the word vector key-value pairs output by the Embedding model, including: inputting the setting requirement data into the vector extraction layer for vector extraction, so as to map each parameter category in the setting requirement data and the corresponding information parameters to the corresponding vector space, and obtaining the vector representations of the information parameters output by the vector extraction layer; inputting the vector identifiers of all information parameters into the feature fusion layer for feature fusion to obtain the word vector key-value pairs output by the feature fusion layer. It should be added that the Embedding model can adopt the bge_large_v1.5 model, and specifically, a suitable model can be selected according to the actual design requirements, and no further limitation is made here.

[0043] In an optional embodiment, before matching the similarity between the word vector key-value pairs and each word vector in the word vector library, it includes: obtaining the corresponding project information based on the preset setting type and constructing the setting project entity of the corresponding setting type; disturbing the parameters of the device to be set based on the preset disturbance conditions to obtain the gain coefficient and oscillation period of the corresponding device to be set; determining the three coefficients of the control system (Proportional Integral Derivative, abbreviated as PID) of the corresponding device to be set according to the gain coefficient and oscillation period of the corresponding device to be set and the preset constant of the device to be set; constructing the entity of the device to be set of the corresponding device to be set according to the gain coefficient and oscillation period of the corresponding device to be set, the three PID coefficients of the corresponding device to be set, and the device information of the device to be set obtained previously; constructing a knowledge graph library according to the relationship between the setting project entities of each setting type and the entities of each device to be set; extracting word vectors from the knowledge graph library and obtaining the word vector library based on the extracted word vectors.

[0044] It should be noted that by obtaining the corresponding project information according to the preset tuning type and then constructing the tuning project entity, all subsequent analysis and processing can be carried out closely around the specific type of tuning project; by disturbing the parameters of the tuning equipment based on the preset disturbance conditions, the gain coefficient and oscillation period are obtained, which provides basic data for the subsequent reasonable setting of PID parameters to achieve precise speed regulation; by determining the three PID coefficients according to the gain coefficient, oscillation period and preset constants, the characteristics of the equipment itself and the established control-related parameter requirements are comprehensively considered, so that the calculated PID The coefficients are more in line with the actual control needs of the equipment, avoiding the problem of poor control effect that may be caused by setting parameters based on experience or general formulas; by constructing a knowledge graph library based on the relationship between the tuning project entities of each tuning type and the entities of each device to be tuned, the scattered and fragmented knowledge about the tuning projects and equipment is effectively integrated, which is convenient for technical personnel to fully and systematically understand the knowledge architecture of the entire tuning field, and provides a clear knowledge context and reference basis for complex tuning work, which helps to quickly locate and use relevant knowledge to solve practical problems; by extracting word vectors from the knowledge graph library to obtain a word vector library, the structured knowledge in the knowledge graph is further converted into a vector form that is convenient for computers to perform semantic analysis and similarity matching, so that in the subsequent similarity matching of word vector key-value pairs with the word vector library, the rich knowledge contained in the knowledge graph can be fully utilized to dig out the semantic associations between different entities and concepts, and find the information that best fits the current tuning needs from a more macro and comprehensive perspective, thereby improving the efficiency and accuracy of the entire information retrieval and utilization, and providing strong support for the accurate determination of PID tuning parameters.

[0045] It should be added that the setting project entity can be shown in the following table:

[0046] It should be added that the project name and project ID are mainly used when maintaining the knowledge graph library or accessing variable presentation, and have no practical significance in the subsequent calculation process. Other attributes will be used as features during subsequent similarity matching. It is worth noting that for the designed cooling capacity, flow rate, head, etc. of the project, the corresponding value ranges can be set based on prior experience or specific design requirements; the thermal region where the project is located can be divided according to the thermal region where the project is located, including severe cold type, cold type, hot summer and cold winter type, hot summer and warm winter type, and mild type, or can be set according to actual design requirements; the project type can also be set according to actual needs, and the project type value can be defined by corresponding preset values for different classification labels. For example, subway is 1, commercial and office building is 2, factory park is 3, etc. The corresponding project type is determined by mapping the corresponding value to the corresponding classification through a hash table; for the longitude and latitude range of the thermal region where the project is located, it can be calibrated by national standards or corrected independently according to actual design requirements, and stored in the form of key-value pairs, where the key is the name of the longitude and latitude, and the value is the upper and lower limits of the longitude and latitude.

[0047] Generally, the above project entity attributes are only examples, and can be modified or added specifically according to the project attributes involved in actual PID tuning. No further limitations are made here. At the same time, if some projects or devices do not have a certain attribute, an empty value placeholder can also be used to ensure the consistency of the vector length and the position of the corresponding attribute during subsequent similarity calculation.

[0048] In addition, based on the preset perturbation conditions, the parameters of the device to be tuned are perturbed to obtain the gain coefficient and oscillation period of the corresponding device to be tuned, including: based on the preset perturbation conditions, the parameters of the device to be tuned are perturbed multiple times to obtain the extreme points of the parameters affected by the perturbation; according to the extreme points of the parameters obtained from each perturbation, the time interval between adjacent maximum or minimum values is determined; according to the time intervals obtained from multiple perturbations, the average value is calculated to determine the oscillation period; the gain value when the device is in a critically stable state (the output signal oscillates with equal amplitude) during the perturbation process is obtained to obtain the gain coefficient.

[0049] The entity of the device to be tuned can be as shown in the following table:

[0050] It should be noted that the preset perturbation conditions can be set based on the specific perturbation requirements of the actual parameters to be perturbed. For example, when the device is a cooling pump, the preset perturbation condition can be to perturb the flow rate. Then, under the preset perturbation conditions, the flow rate of the cooling pump is perturbed, and the flow rate perturbation will affect the temperature parameter. Correspondingly, the extreme point is the temperature extreme point. By obtaining the temperature extreme point, the oscillation period is determined.

[0051] In addition, based on the obtained gain coefficient and oscillation period, and in combination with a preset constant and a target tuning formula, the three PID coefficients corresponding to the equipment to be tuned are determined. The target tuning formula is configured in advance based on actual design requirements and will not be further limited here.

[0052] In addition, the equipment to be tuned can be determined according to the auxiliary equipment of the actual involved system. For example, in a HVAC machine room system, the auxiliary equipment refers to the equipment that plays an auxiliary and supporting role in the operation of the system in the machine room of the HVAC system, excluding the core main equipment (such as chillers, heat pumps, etc.). These auxiliary equipment are crucial for ensuring the efficient, stable and safe operation of the machine room, such as various pumps, water treatment equipment, manifolds, expansion tanks, and various valves.

[0053] It should be noted that the above equipment entities are only examples. The equipment entities can be configured according to the necessary parameters required for the actual design of the PID tuning method and will not be further limited here. Taking the above table as an example, parameters such as the rated power, rated flow, and rated capacity of the equipment can set the corresponding value ranges based on prior experience or specific design requirements; the connection methods and equipment types of multiple devices can be classified with reference to the classification methods of the above project types; the set value, time interval, adjustment step size, gain coefficient Ku, and oscillation period Pu can be determined according to the actual values obtained from the above disturbances; the constants for solving PID and the obtained three PID coefficients can be stored in the form of tuples, lists, or dictionaries.

[0054] The relationship between the project entity and the equipment entity to be tuned can be as follows in the table:

[0055] It should be noted that when extracting word vectors from the knowledge graph library, the same Embedding model as that used for extracting word vectors from the above tuning requirement data should be used to ensure the consistency of word vector extraction between the knowledge graph library and the tuning requirement data, facilitating the subsequent matching of the word vector key-value pairs extracted from the tuning requirement data with each word vector in the word vector library obtained from the knowledge graph library.

[0056] Step S13: Perform similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determine the PID tuning parameters based on the similarity matching results. The word vector library is obtained by extracting word vectors from the previously constructed knowledge graph library, and the knowledge graph library is constructed based on the tuning project entity, the equipment entity to be tuned, and the relationship between the tuning project entity and the equipment entity to be tuned.

[0057] In this embodiment, refer to Figure 3, perform similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determine the PID tuning parameters based on the similarity matching results, including: performing similarity matching between the word vector key-value pairs and each word vector in the word vector library to obtain the similarity matching results, where the similarity matching results include the similarities between the word vector key-value pairs and each word vector in the word vector library; sorting each word vector in the word vector library according to the similarity, and selecting a preset number of word vectors based on the sorting results to obtain the PID tuning parameters.

[0058] It should be noted that by performing similarity matching between the word vector key-value pairs and each word vector in the word vector library, the semantic information contained in the word vectors can be fully utilized to find the word vectors in the word vector library that best match the tuning requirements from the semantic level, avoiding missing key similar experiences or knowledge due to differences in text expressions, and sorting each word vector in the word vector library according to the similarity to arrange the word vectors in an orderly manner according to the degree of relevance to the current requirements, achieving effective screening and focusing of a large amount of information, avoiding the decision-making confusion caused by information overload, and selecting a preset number of word vectors based on the sorting results to preferentially select the word vectors that best meet the requirements, avoiding blindly selecting or randomly intercepting parts of the content among numerous word vectors, extracting the most critical and valuable parts, quickly locking the content with the strongest relevance to the current specific tuning scenario, and thus determining the PID parameters based on these selected information, which helps to improve the efficiency of decision-making and make the entire tuning parameter determination process more efficient and fast.

[0059] In addition, the PID tuning parameters include the gain coefficient, oscillation period, time interval, output step, preset constant, and the three PID coefficients. The form of the PID tuning parameters can be configured correspondingly based on the form of the input tuning requirement data, and no further limitation is made here. For example, the PID tuning parameters can be expressed as: Result: { “tuner_Ku”: 293.78, ”yuner_Pu”: 37.66, ”time_intervel”: 45, “output_step”: 3, ”tuner_divisior”: [330, 270, 336], “tuner_pid_tupte”: [1.78, 0.77, 0.98] } Among them, tuner_Ku represents the gain coefficient, yuner_Pu represents the oscillation period, time_intervel represents the time interval, output_step represents the output step, tuner_divisior represents the preset constant, and tuner_pid_tupte represents the PID three-term coefficient. The specific values of each parameter can be determined based on the specific values corresponding to the word vectors matched in the word vector library determined by the similarity matching result, and no further limitation is made here.

[0060] Furthermore, based on the sorting result, a preset number of word vectors are selected to obtain the PID tuning parameters, including: based on the preset number being 1, according to the sorting result, the word vector with the largest similarity is selected as the PID tuning parameter; based on the preset number being greater than 1, according to the sorting result, in descending order of similarity, a preset number of word vectors are selected and averaged to obtain the PID tuning parameter.

[0061] It should be noted that when the preset number is 1, directly selecting the word vector with the largest similarity as the PID tuning parameter can achieve highly accurate focusing, avoiding interference from other relatively weakly relevant information, making the decision-making process simple, clear, and straightforward; when the preset number is greater than 1, selecting a preset number of word vectors in descending order of similarity and averaging them to obtain the PID tuning parameter, by aggregating the parameters corresponding to these word vectors, to achieve the fusion of multiple relevant information sources, thereby comprehensively considering various situations, avoiding the one-sidedness and limitations that may be brought by relying only on a single piece of information, and making the determined PID parameters more comprehensive and representative.

[0062] In an alternative embodiment, after determining the PID tuning parameters, it includes: based on the user's usage requirements, determining whether to directly use the PID tuning parameters. If so, directly use the PID tuning parameters to configure the corresponding PID controller; otherwise, use the PID tuning parameters for tuning, and optimize the PID tuning parameters based on the tuning result for subsequent use of the optimized PID tuning parameters to configure the corresponding PID controller.

[0063] In summary, in the embodiments of the present invention, word vectors are extracted from the obtained setting requirement data to convert the data in text form into vector representation, so as to better capture the semantic relationships and potential features between words, more accurately extract the key information contained in the setting requirement data, avoid the deviation in the understanding of requirements caused by the surface differences of the text, and further perform similarity matching between the extracted word vector key-value pairs and each word vector in the word vector library, and based on the similarity degree at the semantic level, find the relevant content in the knowledge graph that best matches the setting requirements, and then accurately determine the appropriate PID tuning parameters, making the process of determining PID tuning parameters more scientific and accurate. In addition, a knowledge graph library is constructed based on the setting project entity, the device to be tuned entity, and the relationship between the two, so as to systematically integrate the scattered and fragmented setting knowledge, and further extract a word vector library based on the knowledge graph library, enabling the knowledge in the knowledge graph to be quickly retrieved and matched in vector form, thus facilitating the efficient mining of relevant historical experience and knowledge from the word vector library, quickly locating useful information, improving the efficiency of knowledge utilization, and helping to speed up the entire process of determining PID tuning parameters.

[0064] The PID control system tuning device provided by the present invention will be described below. The PID control system tuning device described below can be correspondingly referred to the PID control system tuning method described above.

[0065] Figure 4 The structural schematic diagram of a PID control system tuning device is shown. The device includes: A data acquisition module 41, which acquires setting requirement data; A word embedding module 42, which extracts word vectors from the setting requirement data to obtain word vector key-value pairs; A parameter determination module 43, which performs similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determines the PID tuning parameters based on the similarity matching result; wherein, the word vector library is obtained by extracting word vectors based on the previously constructed knowledge graph library, and the knowledge graph library is constructed based on the setting project entity, the device to be tuned entity, and the relationship between the setting project entity and the device to be tuned entity.

[0066] The data acquisition module 41 includes: a requirement acquisition unit, which acquires the setting requirements sent by the requester; a requirement reasoning unit, which uses the requirement information prediction model to obtain a requirement reasoning result according to the setting requirements; wherein, the requirement information prediction model is trained based on the requirement training data and the result labels corresponding to the requirement training data; a data integration unit, which extracts keywords from the requirement reasoning result and combines it with a preset format to obtain the setting requirement data.

[0067] Specifically, the demand information prediction model includes a feature extraction layer, a demand prediction layer, an information parameter extraction layer, and an information integration layer, and a demand inference unit, which is used for: inputting demand information into the feature extraction layer for feature extraction to obtain the demand features output by the feature extraction layer; inputting the demand features into the demand prediction layer for demand information prediction to obtain the prediction information output by the demand prediction layer, where the prediction information includes the project information and equipment information required for setting the demand; inputting the demand features into the demand information extraction layer for information parameter extraction to obtain the parameter information output by the demand information extraction layer; and inputting the prediction information and the parameter information into the information integration layer for information integration to obtain the demand inference result output by the information integration layer.

[0068] In addition, the demand inference unit is further used for: inputting the set demand into the demand information prediction model to obtain the demand inference result output by the demand information prediction model; where the demand information prediction model is trained based on the demand training data and the result labels corresponding to the demand training data; determining whether there are missing information parameters in the demand inference result, if there are missing information parameters, generating an information acquisition request based on the missing information parameters and returning it to the demand side to re-acquire the set demand, and inputting the re-acquired set demand into the demand information prediction model to obtain the updated demand inference result output by the demand information prediction model, and re-determining whether there are missing information parameters in the updated demand inference result until there are no missing information parameters in the updated demand inference result.

[0069] The word embedding module 42 is used for: inputting the set demand data into the Embedding model to obtain the word vector key-value pairs output by the Embedding model; where the Embedding model is trained based on the set demand training data and the word vector labels corresponding to the set demand training data.

[0070] Specifically, the Embedding model includes a vector extraction layer and a feature fusion layer, and the word embedding module 42 is further used for: inputting the set demand data into the vector extraction layer for vector extraction to map each parameter category in the set demand data to the corresponding vector space to obtain the vector representations of the respective information parameters output by the vector extraction layer; inputting the vector identifiers of all the information parameters into the feature fusion layer for feature fusion to obtain the word vector key-value pairs output by the feature fusion layer. It should be added that the Embedding model can adopt the bge_large_v1.5 model, and specifically, a suitable model can be selected according to the actual design requirements, and no further limitation is made here.

[0071] In an alternative embodiment, the device further includes: a first entity construction module, which, before matching the word vector key-value pairs with each word vector in the word vector library, obtains corresponding item information based on a preset tuning type and constructs a tuning item entity of the corresponding tuning type; a perturbation module, which perturbs the parameters of the device to be tuned based on preset perturbation conditions to obtain the gain coefficient and oscillation period of the corresponding device to be tuned; a numerical determination module, which determines the PID three-term coefficients of the corresponding device to be tuned according to the gain coefficient and oscillation period of the corresponding device to be tuned and the preset constants of the device to be tuned; a second entity construction module, which constructs a device-to-be-tuned entity of the corresponding device to be tuned according to the gain coefficient and oscillation period of the corresponding device to be tuned, the PID three-term coefficients of the corresponding device to be tuned, and the device information of the device to be tuned obtained previously; a knowledge graph construction module, which constructs a knowledge graph library according to the relationship between the tuning item entities of each tuning type and each device-to-be-tuned entity; and a word vector library construction module, which extracts word vectors from the knowledge graph library and obtains a word vector library based on the extracted word vectors.

[0072] It should be added that the perturbation module includes: a perturbation unit, which perturbs the parameters of the device to be tuned multiple times based on preset perturbation conditions to obtain the extreme points of the parameters affected by the perturbation; a data determination unit, which determines the time interval between adjacent maximum or minimum values according to the extreme points of the parameters obtained by each perturbation; a period acquisition unit, which calculates the average value according to the time intervals obtained by multiple perturbations and determines the oscillation period; and a coefficient acquisition unit, which obtains the gain value when the perturbation process is in a critically stable state (the output signal oscillates with equal amplitude) to obtain the gain coefficient.

[0073] The parameter determination module 43 includes: a matching unit, which matches the word vector key-value pairs with each word vector in the word vector library to obtain a similarity matching result, and the similarity matching result includes the similarity between the word vector key-value pairs and each word vector in the word vector library; and a tuning parameter acquisition unit, which sorts each word vector in the word vector library according to the similarity and selects a preset number of word vectors based on the sorting result to obtain the PID tuning parameters.

[0074] Furthermore, the tuning parameter acquisition unit is configured to: when the preset number is 1, select the word vector with the largest similarity as the PID tuning parameter according to the sorting result; when the preset number is greater than 1, select a preset number of word vectors in descending order of similarity according to the sorting result and perform average aggregation to obtain the PID tuning parameters.

[0075] In an alternative embodiment, the apparatus further includes: a control module, which, after determining the PID tuning parameters, determines whether to directly utilize the PID tuning parameters based on the user's usage requirements. If so, it directly configures the corresponding PID controller with the PID tuning parameters; otherwise, it performs tuning using the PID tuning parameters and optimizes the PID tuning parameters based on the tuning results for subsequent configuration of the corresponding PID controller using the optimized PID tuning parameters.

[0076] In summary, in the embodiment of the present invention, the word embedding module extracts word vectors from the tuning requirement data obtained by the data acquisition module to convert the data in text form into vector representation, so as to better capture the semantic relationships and potential features between words, and more accurately extract the key information contained in the tuning requirement data, avoiding deviations in the understanding of requirements caused by surface differences in the text. Further, the parameter determination module performs similarity matching between the extracted word vector key-value pairs and each word vector in the word vector library, and based on the similarity degree at the semantic level, finds the relevant content in the knowledge graph that best matches the tuning requirements, and then accurately determines the appropriate PID tuning parameters, making the process of determining the PID tuning parameters more scientific and accurate. In addition, a knowledge graph library is constructed based on the tuning project entity, the device to be tuned entity, and the relationship between the two, to systematically integrate the scattered and fragmented tuning knowledge, and further extract a word vector library based on the knowledge graph library, enabling the knowledge in the knowledge graph to be quickly retrieved and matched in vector form, thus facilitating the efficient mining of relevant historical experience and knowledge from the word vector library, quickly locating useful information, improving the efficiency of knowledge utilization, and helping to speed up the entire process of determining the PID tuning parameters.

[0077] Figure 5 An example of the entity structure diagram of an electronic device is shown in Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540. Among them, the processor 510, the communication interface 520, and the memory 530 complete communication with each other through the communication bus 540. The processor 510 can call the logical instructions in the memory 530 to execute the PID control system tuning method, which includes: obtaining tuning requirement data; extracting word vectors from the tuning requirement data to obtain word vector key-value pairs; performing similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determining the PID tuning parameters based on the similarity matching result; where the word vector library is obtained by extracting word vectors based on the previously constructed knowledge graph library, and the knowledge graph library is constructed based on the tuning project entity, the device to be tuned entity, and the relationship between the tuning project entity and the device to be tuned entity.

[0078] In addition, when the logic instructions in the above-mentioned memory 530 can be implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0079] On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the PID control system tuning method provided by the above-mentioned various methods. The method includes: obtaining tuning requirement data; extracting word vector key-value pairs from the tuning requirement data; performing similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determining the PID tuning parameters based on the similarity matching result; wherein the word vector library is obtained by extracting word vectors based on a pre-constructed knowledge graph library, and the knowledge graph library is constructed based on the tuning project entity, the device to be tuned entity, and the relationship between the tuning project entity and the device to be tuned entity.

[0080] On another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the PID control system tuning method provided by the above-mentioned various methods. The method includes: obtaining tuning requirement data; extracting word vector key-value pairs from the tuning requirement data; performing similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determining the PID tuning parameters based on the similarity matching result; wherein the word vector library is obtained by extracting word vectors based on a pre-constructed knowledge graph library, and the knowledge graph library is constructed based on the tuning project entity, the device to be tuned entity, and the relationship between the tuning project entity and the device to be tuned entity.

[0081] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative work.

[0082] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0083] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for tuning a PID control system, characterized in that, Including: Obtain setting requirement data; Extract word vectors from the setting requirement data to obtain word vector key-value pairs; Perform similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determine the PID tuning parameters based on the similarity matching result; wherein, the word vector library is obtained by extracting word vectors from a pre-constructed knowledge graph library, and the knowledge graph library is constructed based on the setting project entity, the device to be tuned entity, and the relationship between the setting project entity and the device to be tuned entity.

2. The PID control system tuning method according to claim 1, characterized in that Performing similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determining the PID tuning parameters based on the similarity matching result, including: Perform similarity matching between the word vector key-value pairs and each word vector in the word vector library to obtain a similarity matching result, where the similarity matching result includes the similarity between the word vector key-value pairs and each word vector in the word vector library; Sort each word vector in the word vector library according to the similarity, and select a preset number of word vectors based on the sorting result to obtain the PID tuning parameters.

3. The PID control system tuning method according to claim 2, characterized in that The selecting a preset number of word vectors based on the sorting result to obtain the PID tuning parameters includes: Based on the preset number being 1, select the word vector with the largest similarity as the PID tuning parameter according to the sorting result; Based on the preset number being greater than 1, select the preset number of word vectors in descending order of similarity according to the sorting result and perform average aggregation to obtain the PID tuning parameters.

4. The PID control system tuning method according to claim 1, characterized in that, The extracting word vectors from the setting requirement data to obtain word vector key-value pairs includes: Input the setting requirement data into an embedded Embedding model to obtain the word vector key-value pairs output by the Embedding model; wherein, the Embedding model is trained based on the setting requirement training data and the corresponding word vector labels of the setting requirement training data.

5. The PID control system tuning method according to claim 1, wherein The obtaining setting requirement data includes: Obtain the setting requirements sent by the requester; According to the setting requirements, use the demand information prediction model to obtain a demand inference result; wherein, the demand information prediction model is trained based on the demand training data and the corresponding result labels of the demand training data; Extract keywords from the demand inference result and combine them with a preset format to obtain the setting requirement data.

6. The PID control system tuning method according to claim 5, wherein According to the setting requirements, using the demand information prediction model to obtain a demand inference result, including: Input the setting requirements into the demand information prediction model to obtain the demand inference result output by the demand information prediction model; wherein, the demand information prediction model is trained based on the demand training data and the corresponding result labels of the demand training data. Determine whether there are missing information parameters in the demand reasoning result. If there are missing information parameters, generate an information acquisition request based on the missing information parameters and return it to the demand side to re-acquire the setting demand, and input the re-acquired setting demand into the demand information prediction model to obtain the updated demand reasoning result output by the demand information prediction model, and re-determine whether there are missing information parameters in the updated demand reasoning result until there are no missing information parameters in the updated demand reasoning result.

7. The PID control system tuning method according to any one of claims 1-6, characterized in that, Before performing similarity matching between the word vector key-value pairs and each word vector in the word vector library, it includes: Based on the preset setting type, obtain the corresponding project information and construct a setting project entity of the corresponding setting type. Based on the preset disturbance conditions, disturb the parameters of the device to be set to obtain the gain coefficient and oscillation period of the corresponding device to be set. According to the gain coefficient and oscillation period of the corresponding device to be set and the preset constant of the device to be set, determine the PID three-term coefficients of the corresponding device to be set. According to the gain coefficient and oscillation period of the corresponding device to be set, the PID three-term coefficients of the corresponding device to be set, and the device information of the device to be set obtained previously, construct a device entity to be set of the corresponding device to be set. Construct a knowledge graph library according to the relationship between the setting project entities of each setting type and each device entity to be set. Extract word vectors from the knowledge graph library and obtain a word vector library based on the extracted word vectors.

8. A PID control system tuning device, characterized in that, It includes: A data acquisition module that acquires setting demand data. A word embedding module that extracts word vectors from the setting demand data to obtain word vector key-value pairs. A parameter determination module that performs similarity matching between the word vector key-value pairs and each word vector in the word vector library, and determines the PID setting parameters based on the similarity matching result; wherein, the word vector library is obtained by previously extracting word vectors based on the constructed knowledge graph library, and the knowledge graph library is constructed based on the setting project entity, the device entity to be set, and the relationship between the setting project entity and the device entity to be set.

9. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the PID control system tuning method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the PID control system tuning method according to any one of claims 1 to 7.

Images