Searching connection system and method for industrial instrument

Abstract

The invention relates to the field of instrument lookup and connection, in particular to a lookup and connection system and method for an industrial instrument, and the system comprises an instrument data collection module, an intelligent matching module, a connection management module and a topological optimization module. The acquisition module completes authentication through zero-trust handshake, optimizes a process in combination with reinforcement learning, and acquires communication, signal and environment characteristics; the intelligent matching module adopts an improved weighted greedy algorithm for rapid matching, and starts an improved dynamic time warping algorithm for secondary deep matching when the threshold value is lower than the threshold value; the connection management module generates candidate schemes, evaluates quality based on LSTM and a random forest model, and solves a Nash equilibrium screening optimal scheme in combination with a game theory; the topological optimization module is fused with multi-source positioning data to construct a three-dimensional space relation matrix, and an optimized communication topological graph is constructed through an improved Delaunay algorithm with signal-to-noise ratio constraint. According to the invention, the connection safety, the matching accuracy, the stability and the topology adaptability of the industrial instrument are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of instrument search connection, in particular to a search connection system and method for industrial instruments. BACKGROUND

[0002] In the field of industrial automation, as the core equipment for monitoring and controlling the production process, the stable and safe connection of industrial instruments is the key to ensuring the continuity of production and the reliability of data. The traditional search connection method for industrial instruments has many technical pain points:

[0003] The types of industrial field devices are diverse, the communication protocols are complex, and the device interface specifications are different, which leads to low efficiency of instrument identification and matching, and easy to misjudge the model or incompatible protocols, especially for similar model instruments. The complexity of the industrial environment, device shielding and distance difference seriously affect the communication quality. The traditional connection scheme relies on fixed parameter configuration and lacks dynamic evaluation and optimization capability for connection quality, often resulting in unstable connection, high packet loss rate or delay exceeding the standard. In the industrial scene, the safety requirement is strict, but the traditional connection authentication mechanism is simple and lacks strict identity verification and encryption transmission process, which poses security risks such as device cloning and data leakage.

[0004] Therefore, a search connection system and method for industrial instruments are needed to solve the above problems. SUMMARY

[0005] In order to solve the technical problems proposed in the background art, the present application provides a search connection system and method for industrial instruments.

[0006] The purpose of the present application can be achieved by the following technical solutions:

[0007] The first aspect of the present application provides a search connection system for industrial instruments, comprising an instrument data acquisition module, an intelligent matching module, a connection management module, a topology optimization module and an industrial instrument feature library, specifically:

[0008] The instrument data acquisition module scans the interface to form a list of instruments to be connected after self-checking and initialization, triggers a zero-trust handshake, verifies the certificate and signature, and issues a token after passing. The instrument with the token collects signals, filters and processes, obtains communication, signal and environmental feature parameters, and combines with the reinforcement learning optimization process. The specific process is as follows:

[0009] Self-check the working state of all interfaces and sensors, and load the preset acquisition parameters such as sampling frequency and data format, initialize the data buffer, establish the communication connection with the system main controller; activate each communication interface in turn, scan according to the preset address range, send a general detection instruction to each instrument address, detect whether there is a responsive instrument to be connected, record all detected instrument addresses and corresponding interface types, and form a list of instruments to be connected;

[0010] When the list of instruments to be connected is formed, the zero-trust handshake protocol process is triggered. Specifically, the instrument to be connected generates a temporary key pair and a detection message, the detection message includes a temporary public key, an instrument certificate and a current timestamp of the device, and sends the temporary key pair and the detection message to the system main controller. The system main controller detects whether the instrument certificate signature is valid and the certificate revocation status. If the detection is invalid or revoked, return error code 401 and perform the reinforcement learning reward step for optimization. If the detection is normal, generate a random challenge code, encrypt the random challenge code through the temporary public key, and return the challenge code to the instrument to be connected. The instrument to be connected decrypts the challenge code through the temporary key pair, generates a signature and sends it to the main controller. The main controller verifies the signature with the public key. If the signature verification fails, return error code 402. On the contrary, if the signature is passed, issue a short-term token. All data acquisition communications need to carry the token. If the token expires, return error code 403 and require re-handshake.

[0011] Introduce the reinforcement learning reward function R t The calculation logic of the reward function is: Wherein is the search efficiency term, defined as the total time from the start of device scanning to the completion of zero-trust handshake and the successful establishment of data transmission channel; PT is the collision penalty term, which is obtained by the ratio of the number of collisions per unit time to the total transmission times, μ is the collision penalty weight, Se is the security penalty term, which is 1 when error code 401 or error code 402 is detected, and 0 otherwise. is the security penalty weight, which is constantly optimized through the reinforcement learning reward function.

[0012] Data acquisition is performed on the instrument to be connected with the token. Multiple groups of signal samples are continuously acquired. Sliding window filtering is used to reduce noise of the original signal. The signal strength and signal-to-noise ratio of each instrument to be connected are calculated. The signal type and signal range are identified. The signal type includes analog and digital; The communication parameters and the function code values actually sent or received during communication are recorded and sorted into a function code set supported by the device. The communication parameters include baud rate, data bits and communication response time; The environmental temperature and humidity of the instrument to be connected are measured in real time. The obtained data is divided into communication characteristic parameters, signal characteristic parameters and environmental characteristic parameters.

[0013] The intelligent matching module improves the weighted greedy algorithm, and quickly matches and calculates the comprehensive similarity based on the multi-dimensional feature vector and the feature library. If the comprehensive similarity is lower than the threshold, the improved DTW algorithm is started for secondary identification, and deep matching is performed through time sequence feature fine alignment. The specific steps are as follows:

[0014] The analytic hierarchy process is used to dynamically allocate the feature weight combined with real-time feedback. The weight is updated in real time through the historical matching accuracy, and the update logic is as follows: Wherein rc i,n is the weight of the ith feature in the nth iteration, i is the feature number, is the learning rate, A i is the historical matching accuracy of the ith feature, A * is the average accuracy;

[0015] The similarity of each feature is calculated, and then the comprehensive similarity is obtained by weighting. Specifically, the similarity of the communication feature parameters is calculated by the Euclidean distance and the Jaccard coefficient. The Euclidean distance calculation logic is as follows: Wherein sim cont is the similarity of continuous communication features, x g is the gth continuous communication feature of the instrument to be connected, is the gth continuous communication feature of the standard instrument in the industrial instrument feature library, m is the total number of continuous features, R max is the maximum possible value range of this feature; the Jaccard coefficient calculation logic is as follows: Wherein sim set is the similarity of set-type communication features, SR is the function code supported by the instrument to be identified, SR * is the function code supported by the standard instrument; the total communication feature similarity is calculated by weighting the set-type communication feature similarity and the continuous communication feature similarity, and the signal feature similarity sim signal is calculated by the cosine similarity algorithm, and the calculation logic is as follows: Wherein m2 is the total number of signal feature parameters, g2 is the number of signal feature parameters, is the gth signal feature parameter of the instrument to be identified, is the gth signal feature parameter of the standard instrument to be identified; the environment feature similarity sim env is obtained by the Euclidean distance complement algorithm, and the total communication feature similarity, the signal feature similarity and the environment feature similarity are weighted to obtain the instrument comprehensive similarity S.

[0016] The industrial instrument feature library is used to store the standardized feature information of various known instruments and various types of instruments.

[0017] A local optimal greedy strategy is used to quickly screen the matching results, the candidate instrument types with a similarity greater than a preset communication threshold are screened out from the feature library, the comprehensive similarity of the instrument to be connected and each candidate type is calculated, the largest candidate type is selected as the instrument type matching result, and the similarity of the k features with the highest weight in the matching process is recorded;

[0018] When the maximum comprehensive similarity of the output is less than the preset similarity threshold, a secondary recognition mechanism is triggered, the fine alignment of the feature sequence is performed through the improved dynamic time warping algorithm DTW, the differentiation problem of similar models of instruments is solved, a distance matrix with weight is constructed, the distance matrix DT is used to measure the local difference between each point in the to-be-identified sequence QC and the standard sequence CY, the feature weight is introduced after improvement, the to-be-identified sequence is the instrument time sequence feature obtained by arranging the collected instrument data in time sequence, the standard sequence is the time sequence feature of the j standard instruments closest to the first matching result in the industrial instrument feature library, and the calculation logic is: Where f1 is the number of the to-be-identified sequence, f2 is the number of the standard sequence, d is the feature dimension number, e is the total number of feature dimensions, rc d is the weight of the dth feature;

[0019] A first cumulative distance matrix is calculated based on the distance matrix, the cumulative distance matrix Acc is used to record the minimum cumulative distance from the starting point to the point [f1][f2], so as to obtain the second cumulative distance matrix Acc[l-1] of the terminal point [l-1] of the sequence, l is the total length of the sequence, and the second cumulative distance matrix reflects the overall matching difference. The cumulative similarity sim dtw is obtained by normalizing the cumulative distance matrix, and the calculation logic is:

[0020] max(Acc) is the maximum value in the cumulative distance matrix;

[0021] The cumulative similarity in the standard sequence is calculated respectively, and the maximum value of the cumulative similarity is selected as the secondary type matching result. The instrument type classification includes pressure transmitters, temperature sensors, valve controllers and PLC controllers.

[0022] The connection management module generates a plurality of candidate connection schemes based on the type matching result, evaluates the connection quality of each candidate scheme by using a connection quality prediction model, and constructs a revenue function by taking the candidate connection scheme as a strategy set and the instrument as a game party through a game theory algorithm, the connection quality, resource conflict and energy consumption are fused, the Nash equilibrium is solved through strategy iteration, and the globally optimal instrument scheme is screened out. The specific process is:

[0023] Based on the type matching result, corresponding instrument parameters, different interface configurations and different protocols in the industrial instrument feature library are extracted as scheme content, the instrument parameters include baud rate position, maximum transmission rate, supported data packet size, power consumption and maximum communication distance, the interface configuration includes physical interface type, and the protocol includes LoRaWAN, BLE and HART; various candidate connection schemes are obtained by randomly combining each instrument parameter, different interface configuration and different protocol;

[0024] The LSTM and the random forest are combined to construct a connection quality model, the random forest establishes several decision trees, the Gini coefficient is used for split feature selection, and the training data is the triple data of devices, schemes and quality in historical connection data; for each candidate scheme, the random forest extracts the corresponding features of the scheme, outputs the initial prediction value through the regression algorithm, the prediction value includes the connection success rate, the delay and the packet loss rate, the success rate is predicted by the classification tree success or failure probability, which is converted into percentage, and the delay and the packet loss rate are predicted by the regression tree to obtain its continuous value; the input of the LSTM is the content of the candidate connection scheme, and the output layer also outputs the connection success rate, the delay and the packet loss rate; the success rate, the delay and the packet loss rate of the LSTM and the random forest are respectively weighted and summed to obtain the quality prediction value of the corresponding candidate connection scheme;

[0025] Taking the device as the game participant, a multi-dimensional income function is constructed, and the calculation logic is: YGH(s v ,s -v )=γ v ·QE v -θ·CY v -Δη·ES, wherein YGH(s v ,s -v ) is the income value of the instrument v, which comprehensively reflects the utility when the candidate connection scheme s is selected, s v is the connection scheme of the instrument v, s -v is the scheme combination of all other devices outside the instrument v; γ v is the priority coefficient of the instrument v, QE v is the quality prediction value, θ is the conflict cost coefficient, Δη is the energy consumption cost coefficient, and CY v is the resource conflict cost;

[0026] Each instrument randomly selects a candidate scheme as its initial strategy. The strategy combinations of all instruments are recorded. Based on the initial strategy, the initial revenue of each device is calculated by substituting it into the revenue function. The devices iterate and update their strategies sequentially. Instrument v is fixed while the strategies of other instruments remain unchanged. It traverses its own strategy set and selects a new strategy that maximizes its own revenue. The above steps are repeated until the strategies of all devices remain unchanged for a preset number of iterations. This strategy combination is marked as a Nash equilibrium. For the strategy combinations in the Nash equilibrium state, the final revenue of each instrument is calculated using the revenue function, and the total system revenue is calculated. The strategy combination with the largest total revenue is selected as the optimal instrument scheme. The optimal instrument scheme corresponding to each instrument is extracted as the final connection method.

[0027] The topology construction module fuses multi-source positioning data from the instrument to establish a three-dimensional spatial relationship matrix. Then, it constructs a communication topology map using an improved Delaunay algorithm with signal-to-noise ratio constraints. After SNR filtering and local real-time optimization of the communication topology, the specific steps are as follows:

[0028] By deploying multiple ultra-wideband anchor points in the industrial field, the three-dimensional coordinates of each instrument to be connected are calculated using the time difference of arrival algorithm to obtain the broadband coordinates; then, the Bluetooth coordinates of the instrument are obtained using a Bluetooth gateway that supports angle of arrival technology and a received signal strength indicator.

[0029] The corresponding signal-to-noise ratios are obtained based on broadband coordinates and Bluetooth coordinates. The signal-to-noise ratios are then normalized to obtain the broadband normalized signal-to-noise ratio FC and the Bluetooth normalized signal-to-noise ratio MS, respectively.

[0030] Based on the broadband normalized signal-to-noise ratio, the broadband coordinate fusion weight Rc is obtained through boundary constraints. Its calculation logic is: Rc = max(0.3, min(0.9, FC));

[0031] Based on the broadband coordinate fusion weight, the broadband coordinates and corresponding Bluetooth coordinates are weighted and summed to obtain the fused coordinates (gx, gy, gz) of each instrument. A spatial relationship matrix is ​​constructed from the fused coordinates of each instrument. The matrix parameters of the spatial relationship matrix include the distance between devices, the signal-to-noise ratio of inter-device communication, and the azimuth angle between devices. The matrix parameters, D, are calculated based on the fused coordinates. v1v2 The formula is: Among them (gx) v1 ,gy v1 ,gz v1 ) and (gx v2 ,gy v2 ,gz v2 The coordinates of instrument v1 and instrument v2 are fused. The signal-to-noise ratio of communication between devices can be directly obtained by acquiring the quality of the communication signals between them. The formula for calculating the azimuth angle between devices is: Where arctan is the arctangent function;

[0032] Taking the instrument fusion coordinates in the spatial relationship matrix as a vertex set, an initial Delaunay triangulation is performed by using a divide-and-conquer algorithm: the vertex set is sorted according to spatial coordinates, and is recursively divided into sub-regions until each sub-region contains ≤ a preset number of vertices, an initial triangular face is constructed for each sub-region, and then the sub-regions are merged and the triangular mesh is optimized by using a Lawson algorithm to generate an initial triangular face set, each triangular face being composed of 3 vertices; the signal-to-noise ratio values of the three edges of each triangular face are extracted from the spatial relationship matrix, and the average signal-to-noise ratio is obtained by averaging the signal-to-noise ratio values, and the triangular face with an average signal-to-noise ratio lower than a threshold value is removed;

[0033] Taking the edges in the triangular mesh as candidate connections, the cost of an edge is set as SNR is the signal-to-noise ratio, the closer the distance and the higher the signal-to-noise ratio, the smaller the cost, a Kruskal algorithm is used to construct a minimum spanning tree, all edges are sorted in ascending order of cost, and the edges are added one by one while avoiding forming a loop, until all instrument nodes are connected, and a minimum spanning tree edge set is generated; all edge sets of the instruments are integrated, and a communication topology graph is output, the communication topology graph including edge attributes and topology indexes; the edge attributes include start and end node IDs, SNR values and distances, and the topology indexes include average edge SNR, node average connectivity and maximum connected distance, and the instrument communication is optimized through the communication topology graph.

[0034] Please refer to Figure 2 The second aspect of the present application provides a connection searching method for industrial instruments, and the specific steps are as follows:

[0035] Step one, after system initialization, a scanning is performed to form a list to be connected, and a zero-trust handshake is triggered: a device sends a message containing a temporary public key and a certificate, a controller verifies the certificate, and if the verification is passed, an encrypted challenge code is sent, the device decrypts the signature, and if the verification is passed, a short-term token is obtained, a reinforcement learning reward function is introduced to optimize the process, data of the device with the token is collected, and after filtering processing, signal, communication and environmental characteristic parameters are obtained;

[0036] Step two, by improving a weighted greedy algorithm, a multi-dimensional feature vector is matched with a feature library based on the improved weighted greedy algorithm, and a comprehensive similarity is calculated, if the comprehensive similarity is lower than a threshold value, a secondary identification is started to improve a DTW algorithm, and deep matching is performed through fine alignment of time sequence characteristics;

[0037] Step three, based on the type matching result, a plurality of candidate connection schemes are generated, a connection quality prediction model is used to evaluate the connection quality of each candidate scheme, and a game theory algorithm is used to take the candidate connection schemes as a strategy set, take the instruments as game parties, construct a revenue function integrating connection quality, resource conflict and energy consumption, solve the Nash equilibrium through strategy iteration, and screen out a globally optimal instrument scheme;

[0038] Step four, according to the fusion of multi-source positioning data of the instrument, a three-dimensional space relationship matrix is established, and an improved Delaunay algorithm with signal-to-noise ratio constraint is used to construct the instrument topology graph, and the SNR screening and local real-time optimization communication topology are carried out.

[0039] Compared with the prior art, the beneficial effects of the present application are:

[0040] Improve the accuracy of instrument identification and matching: an improved weighted greedy algorithm is used to realize fast matching of multi-dimensional features, and the feature weight is dynamically allocated by the analytic hierarchy process to ensure the dominant role of key features in matching. For low similarity scenarios, an improved dynamic time warping algorithm is started, and deep matching is realized through fine alignment of time sequence features, effectively solving the difficulty in distinguishing similar models of instruments, and significantly improving the accuracy and robustness of type identification.

[0041] Optimize connection quality and resource allocation: a connection quality prediction mechanism is constructed based on the fusion model of LSTM and random forest, and the performance of the candidate scheme is accurately evaluated. Combined with game theory algorithm, the candidate scheme is taken as the strategy set to construct a revenue function that integrates connection quality, resource conflict and energy consumption, and the Nash equilibrium is solved through strategy iteration to realize the selection of the globally optimal connection scheme, reduce resource competition conflict, reduce energy consumption, and improve the stability and efficiency of multi-device concurrent connection. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description, the following drawings are not deliberately drawn according to the actual size and proportion, and the emphasis is on showing the main idea of the present application.

[0043] Figure 1 The module connection block diagram of the present application.

[0044] Figure 2 The method step diagram of the present application. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings, and obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor also belong to the scope of protection of the present application.

[0046] Please refer to Figure 1 The first aspect of the present application provides a search connection system for industrial instruments, which comprises an instrument data acquisition module, an intelligent matching module, a connection management module, a topology optimization module and an industrial instrument feature library, and the specific working process of each module is:

[0047] The instrument data acquisition module scans the interface after self-checking initialization to form a list of instruments to be connected, triggers a zero-trust handshake, verifies the certificate and signature, and issues a token after passing; the instrument with the token collects signals after filtering processing, obtains communication, signal and environmental characteristic parameters, and combines with the reinforcement learning optimization process, the specific process is as follows:

[0048] Self-check the working state of all interfaces and sensors, and load the preset acquisition parameters such as sampling frequency and data format, initialize the data buffer, and establish a communication connection with the system main controller; activate each communication interface in turn, scan according to the preset address range, send a general detection command to each instrument address, detect whether there is a responsive instrument to be connected, record all detected instrument addresses and corresponding interface types, and form a list of instruments to be connected;

[0049] When the list of instruments to be connected is formed, the zero-trust handshake protocol process is triggered, specifically, the instrument to be connected generates a temporary key pair and a detection message, the detection message includes a temporary public key, an instrument certificate and a current timestamp of the device, and sends the temporary key pair and the detection message to the system main controller, the system main controller detects whether the instrument certificate signature is valid and the certificate revocation status, if the detection is invalid or revoked, returns error code 401 and executes the reinforcement learning reward step for optimization, if the detection is normal, generates a random challenge code, encrypts the random challenge code with the temporary public key, and returns the challenge code to the instrument to be connected; the instrument to be connected decrypts the challenge code with the temporary key pair, generates a signature and sends it to the main controller, the main controller verifies the signature with the public key, if the signature verification fails, returns error code 402, otherwise, if the signature is passed, issues a short-term token, all data acquisition communications need to carry the token, if the token expires, returns error code 403, and requires re-handshake;

[0050] To realize the self-optimization capability of the system, a reinforcement learning reward function R t is introduced, which quantifies the performance of each device connection process to guide the system to dynamically adjust the plan, such as scan order, communication parameters and security verification strength, and finally realizes the optimization goal of efficient search, low conflict and high security, executes the learning reward reinforcement step, and the calculation logic of the reward function is: Wherein is the search efficiency term, defined as the total time from the start of device scanning to the completion of zero-trust handshake and the successful establishment of data transmission channel; PT is the collision penalty term, which is obtained by the ratio of the number of collisions per unit time to the total transmission times, μ is the collision penalty weight, set to 0.42, Se is the security penalty term, which is 1 when error code 401 or error code 402 is detected, otherwise it is 0 when it is normal; is the security penalty weight, set to a fixed value of 0.58, which follows the principle of safety first, and continuously optimizes the zero-trust handshake protocol process through the reinforcement learning reward function.

[0051] Data acquisition is performed on the token-carrying instrument to be connected, a plurality of groups of signal samples are continuously acquired, a sliding window filter is used to perform noise reduction processing on the original signal, the signal strength and signal-to-noise ratio of each instrument to be connected are calculated, the signal type and signal range are identified, the signal type includes analog and digital; the communication parameters and the function code values actually sent or received in the communication process are recorded and sorted into a function code set supported by the device, the communication parameters include baud rate, data bits and communication response time; the environmental temperature and humidity of the instrument to be connected are measured in real time, the possible influence of the environment on the device communication is evaluated, and the obtained data is divided into communication characteristic parameters, signal characteristic parameters and environmental characteristic parameters.

[0052] The intelligent matching module quickly matches and calculates the comprehensive similarity based on the multi-dimensional feature vector and the feature library by improving the weighted greedy algorithm. If the comprehensive similarity is lower than the threshold, secondary identification is started to improve the DTW algorithm for deep matching through fine alignment of the timing characteristics. The specific steps are as follows:

[0053] The feature weight is dynamically allocated by using the analytic hierarchy process combined with real-time feedback, and the weight is updated in real time through the historical matching accuracy. The update logic is as follows: wherein rc i,n is the weight of the ith feature in the nth iteration, i is the feature number, is the learning rate, A i is the historical matching accuracy of the ith feature, A * is the average accuracy; it should be noted that the historical matching accuracy is specifically obtained by adding 1 to the value each time the system successfully completes device matching, and then dividing the current correct matching times by the total matching times to obtain the historical matching accuracy;

[0054] The similarity of each feature is calculated, and then the comprehensive similarity is obtained by weighting. The similarity of the communication characteristic parameters is calculated by the Euclidean distance and the Jaccard coefficient. The Euclidean distance is used to calculate the similarity of continuous values such as baud rate and response delay, and the Jaccard coefficient is used to calculate the similarity of set characteristics such as function codes. The Euclidean distance calculation logic is as follows: wherein sim cont is the similarity of continuous communication characteristics, x g is the gth continuous communication characteristic of the instrument to be connected, is the gth continuous communication characteristic of the standard instrument in the industrial instrument feature library, m is the total number of continuous features, R max is the maximum possible value range of this feature; the Jaccard coefficient calculation logic is as follows: wherein sim set is the similarity of set communication characteristics, SR is the function code supported by the instrument to be identified, SR *is the function code supported by the standard instrument; the total communication feature similarity is obtained by weighted calculation of the set type communication feature similarity and the continuous type communication feature similarity, and the signal feature similarity sim is calculated by the cosine similarity algorithm signal The calculation logic is as follows: Wherein m2 is the total number of signal feature parameters, g2 is the number of signal feature parameters, is the g2th signal feature parameter of the instrument to be identified, is the g2th signal feature parameter of the standard instrument to be identified; the environment feature similarity sim is obtained by the Euclidean distance complement algorithm env The total communication feature similarity, the signal feature similarity and the environment feature similarity are weighted to obtain the instrument comprehensive similarity S.

[0055] The industrial instrument feature library is used to store the standardized feature information of various known instruments and various types of instruments.

[0056] The local optimal greedy strategy is used to quickly screen the matching result, the candidate instrument type with the total communication feature similarity greater than the preset communication threshold is screened from the feature library, the comprehensive similarity of the instrument to be connected and each candidate type is calculated, the maximum candidate type is selected as the instrument type matching result, and the similarity of the k features with the highest weight in the matching process is recorded.

[0057] When the maximum comprehensive similarity of the output is less than the preset similarity threshold, the secondary identification mechanism is triggered, the fine alignment of the feature sequence is performed by the improved dynamic time warping algorithm DTW, the problem of distinguishing similar models is solved, and a distance matrix with weight is constructed. The distance matrix DT is used to measure the local difference between each point in the to-be-identified sequence QC and the standard sequence CY. After improvement, the feature weight is introduced. The to-be-identified sequence is the instrument time sequence feature obtained by arranging the collected instrument data in time sequence, and the standard sequence is the time sequence feature of the j standard instruments closest to the first matching result in the industrial instrument feature library. The calculation logic is as follows: Wherein f1 is the number of the to-be-identified sequence, f2 is the number of the standard sequence, d is the number of the feature dimension, e is the total number of the feature dimension, rc d is the weight of the dth feature dimension. The smaller the DT[f1][f2] is, the more similar the f1 point of QC is to the f2 point of CY.

[0058] A first accumulated distance matrix is calculated based on the distance matrix. The accumulated distance matrix Acc is used to record the minimum accumulated distance from the starting point to the point [f1][f2]. In this way, the second accumulated distance matrix Acc[l-1] of the end point [l-1] of the sequence is obtained. L is the total length of the sequence. The second accumulated distance matrix reflects the overall matching difference. The accumulated similarity sim is obtained by normalizing the second accumulated distance matrix. dtw The calculation logic is as follows:

[0059] max(Acc) is the maximum value in the cumulative distance matrix, sim dtw The closer to 1, the more similar the overall trend of the two sequences is;

[0060] Calculate the cumulative similarity in the standard sequence respectively, and select the maximum cumulative similarity as the quadratic type matching result; it should be noted that the matching result is to obtain the type classification of the instrument, and the instrument type classification is, for example, a pressure transmitter, a temperature sensor, a valve controller and a PLC controller;

[0061] The connection management module generates a plurality of candidate connection schemes based on the type matching result, evaluates the connection quality of each candidate scheme by using a connection quality prediction model, and constructs a revenue function integrating connection quality, resource conflict and energy consumption by taking the candidate connection scheme as a strategy set and the instrument as a game party through a game theory algorithm, solves the Nash equilibrium through strategy iteration, and selects the globally optimal instrument scheme, and the specific process is as follows:

[0062] Based on the type matching result, the corresponding instrument parameters, different interface configurations and different protocols in the industrial instrument feature library are extracted as scheme contents, the instrument parameters include baud rate position, maximum transmission rate, supported data packet size, power consumption and maximum communication distance, etc., wherein the baud rate position is divided into long distance, medium distance and short distance, and other parameters are also classified, the interface configuration includes physical interface types such as RS485, Ethernet and wireless network type interfaces and corresponding quantities; the protocol includes LoRaWAN, BLE and HART; the instrument parameters, different interface configurations and different protocols are randomly combined to obtain a plurality of candidate connection schemes;

[0063] LSTM and random forest are combined to construct a connection quality model, a random forest is established, the Gini coefficient is used for splitting feature selection, and the training data is the three tuple data of equipment, scheme and quality in historical connection data, for each candidate scheme, the random forest extracts the corresponding features of the scheme, and outputs the initial prediction value through a regression algorithm, the prediction value includes connection success rate, delay and packet loss rate, the success rate is predicted by a classification tree to be success or failure probability, which is converted into a percentage, and the delay and packet loss rate are predicted by a regression tree to be continuous values; the input of LSTM is the content of the candidate connection scheme, and the output layer also outputs the connection success rate, delay and packet loss rate; the success rates, delays and packet loss rates of LSTM and random forest are respectively weighted and summed to obtain the quality prediction value of the corresponding candidate connection scheme;

[0064] Taking the equipment as a game participant, a multi-dimensional revenue function is constructed, and the calculation logic is: YGH(s v ,s -v )=γ v ·QEv -θ·CY v -Δη·ES, where YGH(s) v ,s -v The benefit value of instrument v is the overall utility of selecting candidate connection scheme s, where s represents the benefit value of instrument v. v For the connection scheme of instrument v, s -v For the combination of all other equipment besides instrument v; γ v QE is the priority coefficient for instrument v. v θ is the predicted quality value, θ is the conflict cost coefficient, adjusting the weight of the impact of resource conflict on revenue, set to 0.43; Δη is the energy consumption cost coefficient, adjusting the weight of the impact of energy consumption on revenue, set to 0.67; CY v Resource conflict cost quantifies the cost of competing for resources with other devices, such as increased costs when channels, bandwidths or time slots overlap, and zero costs when there is no conflict. ES is energy consumption cost, which is positively correlated with transmission power and communication frequency.

[0065] Each instrument randomly selects a candidate scheme as its initial strategy. The strategy combinations of all instruments are recorded. Based on the initial strategy, the initial revenue of each device is calculated by substituting it into the revenue function. The devices iterate and update their strategies sequentially. Instrument v is fixed while the strategies of other instruments remain unchanged. It traverses its own strategy set and selects a new strategy that maximizes its own revenue. The above steps are repeated until the strategies of all devices remain unchanged for a preset number of iterations. This strategy combination is marked as a Nash equilibrium. For the strategy combinations in the Nash equilibrium state, the final revenue of each instrument is calculated using the revenue function, and the total system revenue is calculated. The strategy combination with the largest total revenue is selected as the optimal instrument scheme. The optimal instrument scheme corresponding to each instrument is extracted as the final connection method.

[0066] The topology construction module fuses multi-source positioning data from the instrument to establish a three-dimensional spatial relationship matrix. Then, it constructs a communication topology map using an improved Delaunay algorithm with signal-to-noise ratio constraints. After SNR filtering and local real-time optimization of the communication topology, the specific steps are as follows:

[0067] By deploying multiple ultra-wideband anchor points in the industrial field, the three-dimensional coordinates of each instrument to be connected are calculated using the time difference of arrival algorithm to obtain the broadband coordinates; then, the Bluetooth coordinates of the instrument are obtained using a Bluetooth gateway that supports angle of arrival technology and a received signal strength indicator.

[0068] The corresponding signal-to-noise ratios are obtained based on broadband coordinates and Bluetooth coordinates. The signal-to-noise ratios are then normalized to obtain the broadband normalized signal-to-noise ratio FC and the Bluetooth normalized signal-to-noise ratio MS, respectively.

[0069] Based on the broadband normalized signal-to-noise ratio (SNR), the broadband coordinate fusion weight Rc is obtained through boundary constraints. The calculation logic is: Rc = max(0.3, min(0.9, FC)). It should be noted that the constraint logic is as follows: the lower limit constraint is 0.3: even if the broadband signal quality is extremely poor (FC < 0.3), 30% of the weight is still retained to avoid a sharp drop in accuracy due to complete reliance on Bluetooth; the upper limit constraint is 0.9: even if the broadband signal quality is extremely good (FC > 0.9), only 90% of the weight is allocated, and 10% of the weight is reserved for Bluetooth as redundancy check. When 0.3 ≤ FC ≤ 0.9, Rc is directly equal to FC, and the SNR weight is completely determined by the signal quality.

[0070] Based on the broadband coordinate fusion weight, the broadband coordinates and corresponding Bluetooth coordinates are weighted and summed to obtain the fused coordinates (gx, gy, gz) of each instrument. A spatial relationship matrix is ​​constructed from the fused coordinates of each instrument. The matrix parameters of the spatial relationship matrix include the distance between devices, the signal-to-noise ratio of inter-device communication, and the azimuth angle between devices. The matrix parameters, D, are calculated based on the fused coordinates. v1v2 The formula is: Among them (gx) v1 ,gy v1 ,gz v1 ) and (gx v2 ,gy v2 ,gz v2 The coordinates of instrument v1 and instrument v2 are fused. The signal-to-noise ratio of communication between devices can be directly obtained by acquiring the quality of the communication signals between them. The formula for calculating the azimuth angle between devices is: Where arctan is the arctangent function;

[0071] Using the instrument fusion coordinates in the spatial relationship matrix as the vertex set, a divide-and-conquer algorithm is employed for initial Delaunay triangulation: the vertex set is sorted by spatial coordinates and recursively divided into sub-regions until each sub-region contains ≤ a preset number of vertices. Initial triangular faces are constructed for each sub-region. Then, the Lawson algorithm is used to merge sub-regions and optimize the triangulation, generating an initial set of triangular faces, each consisting of 3 vertices. It should be noted that Delaunay triangulation is an algorithm that transforms a discrete set of points into a triangular mesh. Its core principle is that the circumcircle of any triangle does not contain any other vertices, ensuring that the generated triangles are as uniform as possible. The Lawson algorithm adjusts the triangle structure through edge flipping operations.

[0072] The signal-to-noise ratio (SNR) values ​​of the three sides of each triangle are extracted from the spatial relationship matrix, and the average SNR is calculated by averaging them. Triangles with an average SNR lower than the threshold are then removed.

[0073] Minimum spanning tree construction steps: Using edges from the triangular mesh as candidate connections, set the cost of each edge to... SNR stands for Signal-to-Noise Ratio. The closer the distance and the higher the SNR, the lower the cost. The Kruskal algorithm is used to construct the minimum spanning tree. All edges are sorted in ascending order of cost, and edges are added sequentially while avoiding loops, until all instrument nodes are connected, generating the edge set of the minimum spanning tree. All edge sets of the instruments are integrated, and a communication topology graph is output. This graph optimizes instrument communication and includes edge attributes and topology metrics. Edge attributes include the start and end node IDs, SNR values, and distances. Topology metrics include average edge SNR, average node connectivity, and maximum connected distance. It should be noted that the Kruskal algorithm is a greedy algorithm used to find the minimum spanning tree of a weighted connected graph. Its core step is to select edges in ascending order of weight while avoiding loops, ultimately generating a tree structure that connects all nodes and has the minimum total weight.

[0074] Please refer to Figure 2 As shown, the second aspect of the present invention provides a method for finding and connecting industrial instruments, the specific steps of which are as follows:

[0075] Step 1: After system initialization, a list of devices to be connected is generated by scanning, triggering a zero-trust handshake: The device sends a message containing a temporary public key and certificate. The controller verifies the certificate. If the certificate is verified, an encryption challenge code is sent. The device decrypts the signature. If the verification is successful, a short-term token is obtained. A reinforcement learning reward function is introduced to optimize the process. Data is collected from the device with the token. After filtering, signal, communication and environmental characteristic parameters are obtained.

[0076] Step 2: By improving the weighted greedy algorithm, the multi-dimensional feature vector is quickly matched with the feature library and the comprehensive similarity is calculated. If it is lower than the threshold, the improved DTW algorithm is started for secondary recognition, and deep matching is performed through fine alignment of temporal features.

[0077] Step 3: Generate multiple candidate connection schemes based on the type matching results, evaluate the connection quality of each candidate scheme using a connection quality prediction model, and construct a payoff function that integrates connection quality, resource conflict and energy consumption using a game theory algorithm with the candidate connection schemes as the strategy set and the instrument as the player. Solve the Nash equilibrium through strategy iteration to select the globally optimal instrument scheme.

[0078] Step 4: Based on the fusion of multi-source positioning data of the instrument, a three-dimensional spatial relationship matrix is ​​established. Then, the instrument topology map is constructed by the improved Delaunay algorithm with signal-to-noise ratio constraints. The communication topology is then filtered by SNR and optimized locally in real time.

[0079] The above formulas all retain numerical calculation logic through dimensionless processing. Specific dimensionless processing can be achieved using industrial data preprocessing methods such as standardization and normalization, which will not be elaborated here. The formulas are derived from software simulation optimization based on a large amount of measured data from industrial sites, and can accurately fit actual application scenarios. The preset parameters in the formulas can be flexibly set by those skilled in the art according to the characteristics of the industrial environment (such as electromagnetic interference intensity and equipment density).

[0080] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, they can be presented in whole or in part as a computer program product. The computer program product includes one or more computer instructions or programs, which, after being loaded and executed on a computer, can fully or partially implement the core processes and functions of the present invention, such as instrument data acquisition, intelligent matching, connection optimization, and topology construction. The computer can be a general-purpose computer, an industrial control computer, a computer network device, or other programmable industrial control device.

[0081] Computer instructions may be stored in a computer-readable storage medium or transmitted between storage media, for example, from an industrial control station, server, or data center to a target device via wired (e.g., Ethernet, RS485) or wireless (e.g., LoRa, Bluetooth, microwave) means. The computer-readable storage medium may be any available medium accessible to a computer, or an industrial-grade data storage device (e.g., a server, edge computing node) containing one or more available media. Available media include magnetic media (e.g., industrial hard drives, magnetic tape), optical media (e.g., industrial-grade DVDs), or semiconductor media (e.g., solid-state drives, industrial control chips).

[0082] It should be understood that the process number in each embodiment of the present invention does not represent the execution order. The execution order of each process is determined by its functional logic and is not limited by the implementation process.

[0083] Those skilled in the art will recognize that the modules and algorithm steps described in conjunction with the embodiments of the present invention can be implemented through a combination of electronic hardware, computer software, and hardware. Whether the function is executed in hardware or software depends on the specific requirements of the industrial scenario (such as real-time requirements and cost constraints). Professionals can choose an appropriate implementation method for different application scenarios, but its implementation should not exceed the protection scope of the present invention.

[0084] In the embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other forms. For example, the module division is only a logical functional division, and in actual implementation, they can be integrated or separated, such as integrating the intelligent matching module and the connection management module into an industrial control host, or separating them into independent edge computing nodes. The coupling between modules can be achieved directly or indirectly through interfaces, and can be in other forms specified by electrical, mechanical, or industrial communication protocols.

[0085] The units described as separate components can be physically separated or integrated. The components shown as units can be physical units or distributed nodes. Some or all of the units can be selected to implement the solution of this embodiment according to actual needs.

[0086] If the function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this, the contribution of the technical solution of this invention to the prior art, or the functional module, can be embodied in a software product. This software product is stored in a medium and contains several instructions that cause a computer device (such as an industrial control computer or server) to execute all or part of the steps of the methods described in the embodiments of this invention. The aforementioned storage medium includes industrial storage media capable of storing program code, such as USB flash drives, industrial-grade portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0087] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An industrial instrument lookup and connection system, comprising an instrument data acquisition module, an intelligent matching module, a connection management module, a topology optimization module, and an industrial instrument feature library, characterized in that: The instrument data acquisition module forms a list of instruments to be connected by scanning the status of each instrument and triggers a zero-trust handshake: the device sends a message containing a temporary public key and certificate, the controller verifies the certificate, and if it passes, sends an encrypted challenge code. The device decrypts the signature and obtains a short-term token if the verification is successful. The reinforcement learning reward function is introduced to optimize the process. Data is collected from the device with the token, filtered, and signal, communication and environmental characteristic parameters are obtained. The intelligent matching module improves the weighted greedy algorithm, quickly matches and calculates the comprehensive similarity based on multi-dimensional feature vectors and feature library. If the similarity is lower than the threshold, the second recognition starts the improved dynamic time warping algorithm, and performs deep matching through fine alignment of temporal features. The connection management module generates multiple candidate connection schemes based on the type matching results, evaluates the connection quality of each candidate scheme using a connection quality prediction model, and constructs a payoff function that integrates connection quality, resource conflict and energy consumption using a game theory algorithm with the candidate connection schemes as the strategy set and the instrument as the player. The Nash equilibrium is solved through strategy iteration to select the globally optimal instrument scheme. The topology optimization module fuses multi-source positioning data from the instrument to establish a three-dimensional spatial relationship matrix. Then, it constructs an instrument topology map using an improved Delaunay algorithm with signal-to-noise ratio constraints. The communication topology is then filtered by signal-to-noise ratio and optimized locally in real time.

2. The industrial instrument lookup and connection system according to claim 1, characterized in that, The intelligent matching module, through an improved weighted greedy algorithm, quickly matches and calculates the comprehensive similarity based on multi-dimensional feature vectors and a feature library. The specific steps are as follows: The analytic hierarchy process (AHP) is used in conjunction with real-time feedback to dynamically allocate feature weights, and the weights are updated in real time based on historical matching accuracy. The similarity is calculated for each type of feature, and then the weights are updated to obtain the comprehensive similarity. Specifically, the similarity of communication feature parameters is calculated using Euclidean distance and Jaccard coefficient. The similarity of set-type communication features and continuous-type communication features is weighted to obtain the total communication feature similarity. The similarity of signal features is calculated using the cosine similarity algorithm. The similarity of environmental features is obtained using the Euclidean distance complement algorithm. The comprehensive similarity of the instrument is obtained by weighting the total communication feature similarity, signal feature similarity, and environmental feature similarity. The industrial instrument feature library is used to store standardized feature information of various known instruments and various types of instruments; A greedy strategy with local optima is adopted to quickly filter matching results. Candidate instrument types with a total communication feature similarity greater than a preset communication threshold are selected from the feature library. The comprehensive similarity between the instrument to be connected and each candidate type is calculated. The candidate type with the largest similarity is selected as the instrument type matching result. The similarity of the k features with the highest weights during the matching process is recorded.

3. The industrial instrument lookup and connection system according to claim 2, characterized in that, The intelligent matching module triggers a secondary recognition to start an improved dynamic time warping algorithm, which performs deep matching through fine alignment of temporal features. When the maximum overall similarity of the output is less than the preset similarity threshold, a secondary recognition mechanism is triggered. The improved dynamic time warping algorithm is used to finely align the feature sequences to solve the problem of distinguishing similar instrument models. A weighted distance matrix is ​​constructed. The sequence to be recognized is the time-series feature of the collected instrument data arranged in chronological order. The standard sequence is the time-series feature of the j standard instruments in the industrial instrument feature library that are closest to the first matching result. The first cumulative distance matrix is ​​calculated based on the distance matrix, and then the second cumulative distance matrix of the endpoint of the sequence is obtained. The cumulative similarity is obtained by normalizing the matrix. Calculate the cumulative similarity in the standard sequences respectively, and select the maximum cumulative similarity as the result of the secondary type matching.

4. The industrial instrument lookup and connection system according to claim 1, characterized in that, The topology construction module establishes a three-dimensional spatial relationship matrix by fusing multi-source positioning data from the instrument, and then constructs a communication topology map using an improved Delaunay algorithm with signal-to-noise ratio constraints. The communication topology is then filtered by signal-to-noise ratio and optimized locally in real time. The specific steps are as follows: By deploying multiple ultra-wideband anchor points in the industrial field, the three-dimensional coordinates of each instrument to be connected are calculated using the time difference of arrival algorithm to obtain the broadband coordinates; then, the Bluetooth coordinates of the instrument are obtained using a Bluetooth gateway that supports angle of arrival technology and a received signal strength indicator. The corresponding signal-to-noise ratios are obtained based on broadband coordinates and Bluetooth coordinates. The signal-to-noise ratios are then normalized to obtain the broadband normalized signal-to-noise ratio and the Bluetooth normalized signal-to-noise ratio. Based on the broadband normalized signal-to-noise ratio, the broadband coordinate fusion weight is obtained through boundary constraints. Based on the broadband coordinate fusion weight, the broadband coordinates and the corresponding Bluetooth coordinates are weighted and summed to obtain the fused coordinates of each instrument. The fused coordinates of each instrument are used to construct a spatial relationship matrix. The matrix parameters of the spatial relationship matrix include the distance between devices, the communication signal-to-noise ratio between devices, and the azimuth angle between devices. Using the instrument fusion coordinates in the spatial relationship matrix as the vertex set, a divide-and-conquer algorithm is used to perform the initial Delaunay triangulation: the vertex set is sorted according to spatial coordinates and recursively divided into sub-regions until each sub-region contains less than or equal to a preset number of vertices. An initial triangular face is constructed for each sub-region. Then, the Lawson algorithm is used to merge the sub-regions and optimize the triangulation to generate an initial set of triangular faces, each of which consists of three vertices. The signal-to-noise ratio (SNR) values ​​of the three edges of each triangular face are extracted from the spatial relationship matrix and their average values ​​are calculated to obtain the average SNR. Triangular faces with an average SNR lower than a threshold are removed. Using the edges in the triangular network as candidate connections, the cost of the edge is set by dividing the corresponding distance between devices by the signal-to-noise ratio. The minimum spanning tree is constructed using the Kruskal algorithm. All edges are sorted in ascending order of cost, and edges are added sequentially while avoiding loops, until all instrument nodes are connected, generating the edge set of the minimum spanning tree. All edge sets of the instruments are integrated, and the communication topology graph is output. The communication topology graph is used to optimize the communication of the instruments.

5. The industrial instrument lookup and connection system according to claim 1, characterized in that, The specific process by which the instrument data acquisition module triggers a zero-trust handshake is as follows: The system performs a self-check of the working status of all interfaces and sensors, loads preset acquisition parameters such as sampling frequency and data format, initializes the data buffer, and establishes a communication connection with the system's main controller. It then activates each communication interface in sequence, scans according to the preset address range, sends a general probe command to each instrument address, detects whether there are any responsive instruments to be connected, records all detected instrument addresses and their corresponding interface types, and forms a list of instruments to be connected. When a list of instruments to be connected is formed, the zero-trust handshake protocol process is triggered. A temporary key pair and a detection message are generated for the instruments to be connected. The temporary key pair and the detection message are sent to the system main controller. The system main controller checks whether the instrument certificate signature is valid and the certificate revocation status. If the detection is invalid or revoked, error code 401 is returned and the reinforcement learning reward step is executed for optimization. If the detection is normal, a random challenge code is generated, the random challenge code is encrypted with a temporary public key, and the challenge code is returned to the instruments to be connected. The instrument to be connected decrypts the challenge code using a temporary key pair, generates a signature, and sends it to the main controller. The main controller verifies the signature using the public key. If the signature verification fails, it returns error code 402. Conversely, if the signature passes, it issues a short-term token. All data acquisition communications must carry the token. If the token expires, it returns error code 403, requiring a re-handshake. A reinforcement learning reward function is introduced to optimize the zero-trust handshake protocol process.

6. The industrial instrument lookup and connection system according to claim 5, characterized in that, The instrument data acquisition module acquires communication, signal, and environmental characteristic parameters from the instrument to be connected carrying the token. The specific process is as follows: Data is acquired from the instruments to be connected carrying tokens. Multiple sets of signal samples are continuously acquired. The original signals are denoised using sliding window filtering. The signal strength and signal-to-noise ratio of each instrument to be connected are calculated. The signal type and signal range are identified. The signal type includes analog and digital signals. The communication parameters and the function code values ​​actually sent or received during the communication process are recorded and organized into a set of function codes supported by the device. The communication parameters include baud rate, data bits, and communication response time. Then, the ambient temperature and humidity of the device to be connected are measured in real time, and the obtained data is divided into communication characteristic parameters, signal characteristic parameters and environmental characteristic parameters.

7. The industrial instrument lookup and connection system according to claim 1, characterized in that, The connection management module generates multiple candidate connection schemes based on the type matching results, and evaluates the connection quality of each candidate scheme using a connection quality prediction model. The specific process is as follows: Based on the type matching results, the corresponding instrument parameters, different interface configurations, and different protocols in the industrial instrument feature library are extracted as the solution content; the instrument parameters, different interface configurations, and different protocols are randomly combined to obtain a variety of candidate connection schemes; A connection quality model is constructed by combining LSTM and random forest. The random forest builds several decision trees, and the splitting feature selection adopts the Gini coefficient. The training data is the triple data of device, scheme and quality in the historical connection data. For each candidate scheme, the random forest extracts the corresponding features of the scheme and outputs the initial prediction value through the regression algorithm. The success rate is predicted by the classification tree to predict the probability of success or failure and converted into a percentage. The latency and packet loss rate are predicted by the regression tree to predict their continuous values. The LSTM input consists of candidate connection schemes, and the output layer outputs connection success rate, latency, and packet loss rate. The success rate, latency, and packet loss rate of the LSTM and Random Forest are weighted and summed to obtain the quality prediction value of the corresponding candidate connection scheme.

8. The industrial instrument lookup and connection system according to claim 7, characterized in that, The connection management module uses a game theory algorithm to construct a payoff function with candidate connection schemes as the strategy set and the instrument as the player. It then solves for the Nash equilibrium through strategy iteration to select the globally optimal instrument scheme. By treating devices as game participants, a multi-dimensional payoff function is constructed. Each instrument randomly selects a candidate scheme as its initial strategy. The strategy combinations of all instruments are recorded. Based on the initial strategy, the initial revenue of each device is calculated by substituting it into the revenue function. The devices iterate and update their strategies sequentially. The strategies of other instruments remain unchanged while the instrument itself is fixed. The instrument traverses its own strategy set and selects a new strategy that maximizes its own revenue. The above steps are repeated until the strategies of all devices remain unchanged for a preset number of iterations. This strategy combination is marked as a Nash equilibrium. For the strategy combinations in the Nash equilibrium state, the final revenue of each instrument is calculated using the revenue function, and the total revenue of the system is calculated. The strategy combination with the largest total revenue is selected as the optimal instrument scheme. The optimal instrument scheme corresponding to each instrument is extracted as the final connection method.

9. A lookup and connection method for industrial instruments, used to implement the lookup and connection system for industrial instruments as described in any one of claims 1-8, characterized in that, The specific steps are as follows: Step 1: After system initialization, a list of devices to be connected is generated by scanning, triggering a zero-trust handshake: The device sends a message containing a temporary public key and certificate. The controller verifies the certificate. If the certificate is verified, an encryption challenge code is sent. The device decrypts the signature. If the verification is successful, a short-term token is obtained. A reinforcement learning reward function is introduced to optimize the process. Data is collected from the device with the token. After filtering, signal, communication and environmental characteristic parameters are obtained. Step 2: By improving the weighted greedy algorithm, the multi-dimensional feature vector is quickly matched with the feature library and the comprehensive similarity is calculated. If it is lower than the threshold, the improved DTW algorithm is started for secondary recognition, and deep matching is performed through fine alignment of temporal features. Step 3: Generate multiple candidate connection schemes based on the type matching results, evaluate the connection quality of each candidate scheme using a connection quality prediction model, and construct a payoff function that integrates connection quality, resource conflict and energy consumption using a game theory algorithm with the candidate connection schemes as the strategy set and the instrument as the player. Solve the Nash equilibrium through strategy iteration to select the globally optimal instrument scheme. Step 4: Based on the fusion of multi-source positioning data of the instrument, a three-dimensional spatial relationship matrix is ​​established. Then, the instrument topology map is constructed by the improved Delaunay algorithm with signal-to-noise ratio constraints. The communication topology is then filtered by signal-to-noise ratio and optimized locally in real time.

Images