Local mixed g-good adjacent and ring diagnosis combined fault diagnosis method for tumor medical system

Abstract

The invention discloses a local mixed g-good adjacent and ring diagnosis combined fault diagnosis method for a tumor medical system, and belongs to the technical field of system-level fault diagnosis, and the method comprises the steps: starting from a fault-free state node of the tumor medical system, diagnosing faults of an adjacent node and a communication link according to an HPMC model test rule, and carrying out the fault diagnosis of the adjacent node and the communication link; marking nodes and communication links of which the states cannot be determined; using a local g-good neighbor condition to further diagnose faults of nodes and communication links of which the states cannot be determined, and marking residual nodes and communication links of which the states cannot be determined; and searching a ring structure from adjacent nodes of the remaining nodes of which the states cannot be determined, further diagnosing faults of the remaining nodes by using a ring diagnosis rule, and diagnosing faults of communication links among the remaining nodes in combination with an HPMC model test rule. According to the invention, the fault source of the complex system can be positioned more conveniently, efficiently and accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of system-level fault diagnosis, and more specifically, to a fault diagnosis method and device combining local mixed g-good neighbor and ring diagnosis for tumor medical systems. Background Art

[0002] Fault diagnosis refers to the process of monitoring and analyzing the operating status, parameters and performance of a system to determine the type, location and cause of system failures. It is particularly critical in multi-processor systems, medical equipment and complex network systems, helping the system to detect problems in a timely manner, isolate faulty components and maintain the overall performance of the system when failures occur.

[0003] The tumor medical system is a system that uses artificial intelligence technology to assist doctors in evaluating and diagnosing tumor patients. By analyzing the patient's clinical data, imaging materials and laboratory test results, it helps doctors determine the type, grade, prognosis and other information of the tumor more quickly and accurately, thereby assisting doctors in formulating more effective treatment plans. It integrates digitalization, automation, communication and control technologies, aiming to improve the accuracy and efficiency of patient tumor diagnosis, help patients obtain personalized and targeted treatment plans in a timely manner, and improve treatment effects and survival rates.

[0004] With the development of information technology, the structure of medical systems is becoming increasingly complex, and node and communication link failures will inevitably affect the normal operation of the system. Therefore, it is very necessary to perform system-level fault diagnosis on the medical system and check the status of the processor nodes and communication links connected to it in real time, so as to locate the source of the fault in the system more efficiently and accurately, reduce the impact of the fault on time, money and manpower, and improve the overall stability and reliability of the system.

[0005] In system-level fault diagnosis, fault diagnosis methods based on graph theory models are a common means of achieving system self-diagnosis. Graph theory models define the association between nodes in the system, which is used to evaluate the maximum number of faulty nodes that the system can accommodate and identify faults. With the increasing application of large-scale, multi-node systems in computer technology and medical fields, the complexity of the system is also increasing; during operation, the possibility of failure of each node and its communication link increases, posing a threat to the normal operation of the system.

[0006] For example, a tumor medical system consisting of numerous diagnostic links, treatments, and health monitoring can be viewed as an undirected graph. Each equipment inspection link, diagnostic result, treatment method, and health monitoring process in the tumor medical system can be viewed as a node in the graph. In addition, the inspection results of each link in the tumor medical system, the association between inspections, and the association between diagnostic reports and medication can be viewed as edges in the graph. If a certain inspection node in the tumor medical system fails, the subsequent inspection results and treatment will be affected. According to the speed limit of tumor spread, if the fault source cannot be located in time and the faulty equipment cannot be replaced, the diagnosis result will be incorrect, making it impossible for the patient to receive effective treatment.

[0007] Among the graph theory models for system-level fault diagnosis, the PMC model is a widely used diagnostic model. They usually only allow adjacent processor nodes to perform mutual self-fault diagnosis. Generally, the fault-free processor nodes test the neighboring nodes in turn, and the state of the processor nodes is judged by the test results. If the test result is 0 (or 1), the state of the tested node is diagnosed to be fault-free (or the tested node is faulty). However, the fault diagnosis method based on the PMC model is limited to considering faulty processor nodes, but ignores the importance of communication links. With the increase in the number of large-scale processors, the probability of failure of the communication links connecting processor nodes gradually increases. In 2020, researchers proposed a new hybrid PMC model (HPMC model) based on the PMC model. The HPMC model is the first multi-fault self-diagnosis model for node and communication link failures. The model tests the fault-free nodes to their neighboring nodes and considers the state of the communication edges between nodes at the same time, thereby achieving a more comprehensive diagnosis of the system.

[0008] However, in both the HPMC model and the PMC model, the fault diagnosis process only focuses on the global state of the system and ignores the local state of the system. For example, if system G is formed by integrating two arbitrary subsystems in some way, and the two subsystems can diagnose the maximum number of faulty processors or faulty communication links as m and n, respectively, and m is much larger than n, then the diagnostic capability of system G is limited to n, but even if the number of faulty processors and communication links is between m and n, all faulty processors or faulty communication links can be correctly identified. Therefore, only considering the global fault / no fault state will lose some local system details, which can easily lead to a decrease in the accuracy of system fault diagnosis.

[0009] The Chinese patent with publication number CN118473909A discloses a local hybrid fault self-diagnosis method for a fruit supply chain, which is a prior application of the applicant. Based on the HPMC model, the patent uses a local point-edge hybrid fault diagnosis technology to diagnose the fruit supply chain network, shortening the time for fruit supply chain fault diagnosis and improving the accuracy of detection. However, the diagnostic method is designed for the special structure of the fruit supply chain, while the tumor medical system has a more complex ring structure. Specifically, when there are multiple nodes in the system forming a closed loop, even if a node fails and is removed, the connectivity of the entire system will not be damaged, resulting in the inability to diagnose some node or communication link failures. Summary of the invention

[0010] The purpose of the present invention is to provide a fault diagnosis method for a tumor medical system that combines local hybrid g-good neighbors with ring diagnosis. The HPMC model is used to first diagnose the local state of the system. The state of uncertain nodes can be judged by their neighboring nodes. According to the condition that each fault-free node has at least g (g≥1) good neighbors, the state of each uncertain node is determined in turn, thereby improving the accuracy and reliability of fault diagnosis. At the same time, according to the ring structure characteristics of the tumor medical system, a ring diagnosis method is introduced to improve the system's coverage of local and global faults, and further improve the accuracy and efficiency of diagnosis.

[0011] In order to achieve these objectives of the present invention, the present invention provides a fault diagnosis method combining local hybrid g-good neighbor and loop diagnosis for a tumor medical system, which combines the closed-loop, dynamic and real-time characteristics of the tumor medical system and optimizes the local hybrid g-good neighbor fault diagnosis under the HPMC model by constructing a special structure, specifically including:

[0012] S1. Starting from a node in a non-fault state of the tumor medical system (setting the node to a non-fault state or selecting a node in a non-fault state), according to the HPMC model test rules, diagnosing the neighboring node or communication link failure, and marking the nodes and communication links whose states cannot be determined;

[0013] S2, using the local g-good neighbor condition to further diagnose the failure of nodes and communication links whose states cannot be determined, and marking the remaining nodes and communication links whose states cannot be determined;

[0014] S3. Start searching the ring structure from the adjacent nodes of the remaining nodes whose states cannot be determined, use the ring diagnosis rules to further diagnose the faults of the remaining nodes, and diagnose the faults of the communication links between the remaining nodes in combination with the HPMC model test rules.

[0015] Specifically, the HPMC model test rules of the present invention can be obtained from the prior art (such as Chinese patent publication number CN118473909A).

[0016] Specifically, the local g-good neighbor condition of the present invention is: there are at least g fault-free nodes for each fault-free node, where g is greater than or equal to 1 (g≥1).

[0017] Specifically, the ring structure refers to a ring structure having a number of nodes connected through communication links to form a closed loop. In one embodiment, it includes: node a, node b, node c, node d, node e and node f, which are sequentially connected to form a closed loop through communication links.

[0018] Preferably, the tumor medical system is first divided into several local network structures, and the local network structure is selected to execute the steps S1 to S3 shown, so as to obtain the minimum number of faulty nodes and communication links in the fault diagnosis results of all local network structures, and then, based on the relationship between global and local diagnosis, the maximum number of faults in the entire tumor medical system is obtained.

[0019] Preferably, a fault diagnosis method combining local hybrid g-good neighbor and ring diagnosis for a tumor medical system of the present invention specifically includes:

[0020] Step 1: Construct the network structure of the tumor medical system and divide it into several local network structures, select the initial nodes, and set the initial state;

[0021] Step 2: Starting from the initial node, perform breadth-first traversal to traverse the degree of each node in the tumor medical system network structure to obtain the node degree matrix of the tumor medical system;

[0022] Step 3: Test the network structure of the tumor medical system according to the test rules of the HPMC model, obtain the test symptom set, perform binary encoding, generate a point-edge mixed fault set, and establish the symptom set through state assignment, point-edge triple assignment and diagnostic testing;

[0023] Step 4: Perform a depth-first traversal of the local network structure of the selected tumor medical system to find the same symptoms as those in the symptom set, and quickly perform the first round of fault node and fault edge diagnosis and location, and mark the nodes whose status cannot be determined;

[0024] Step 5: For nodes whose states cannot be determined, start searching from the neighboring nodes of the nodes whose states cannot be determined, and find multiple ring structures respectively according to the degree matrix; for these ring structures containing uncertain node states, further determine the faulty nodes by using the closed characteristics of the rings; after completing the diagnosis of the nodes whose states cannot be determined, determine whether the states of the communication links between these nodes are faulty or not according to the test symptom set of the HPMC model;

[0025] Step 6: Update status;

[0026] Step 7: Repeat steps 1 to 7 to locate the faulty nodes and faulty links in the entire tumor medical system.

[0027] Since the diagnostic capability of the global tumor medical system is limited by the diagnostic capability of the local network structure, the minimum number of faulty nodes and faulty communication edges of the local network structure obtained by traversal is the number of faulty nodes and faulty communication edges of the global tumor medical system, or the maximum diagnostic capability of the global tumor medical system is limited by the minimum number of faulty nodes and faulty communication edges of the local network structure. For example, if system G is formed by integrating two arbitrary subsystems in some way, and the two subsystems can diagnose the maximum number of faulty processors or faulty communication links as m and n respectively, and m is much larger than n, then the diagnostic capability of system G is limited by n. Therefore, even if the number of faulty processors or faulty communication links is between m and n, system G can only correctly identify no more than n faulty processors or faulty communication links.

[0028] The present invention provides a computing device, comprising a processor and a memory, wherein the memory stores executable codes, and when the processor executes the executable codes, the local mixed fault diagnosis method is implemented.

[0029] The present invention has at least the following beneficial effects:

[0030] The present invention first divides the tumor medical system into multiple local structures, and uses the test rules of the HPMC model to determine the status of each node and communication link in the local structure. For nodes with uncertain status, judgment is made in combination with their neighboring nodes and g-good neighbor conditions. At the same time, the ring diagnosis method ensures that in a complex topology containing a ring structure, even if some local areas are divided due to faults, the system can still accurately identify the faulty nodes and communication links.

[0031] The present invention introduces ring diagnosis into the fault diagnosis of tumor medical systems, significantly improving the fault detection accuracy of the system at the local and global levels. Based on the relationship between local and global diagnostic results, the system can effectively obtain the maximum number of faulty nodes and faulty communication edges to ensure the comprehensiveness and accuracy of the diagnostic results. This method greatly improves the diagnostic efficiency and enhances the system's comprehensive evaluation and fault isolation capabilities, and is particularly suitable for high-reliability scenarios such as tumor medical systems.

[0032] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a directed graph of the network structure of a tumor medical system under the HPMC model of the present invention;

[0034] Figure 2 It is an example of local g-good neighbor condition fault diagnosis under the HPMC model of the present invention;

[0035] Figure 3 It is the fault diagnosis process of the "little diamond" structure constructed by the present invention;

[0036] Figure 4 It is the state transfer process of the present invention;

[0037] Figure 5 It is the breadth-first traversal process of the present invention;

[0038] Figure 6 It is the process of constructing a symptom set of the present invention;

[0039] Figure 7 It is the depth-first traversal process of the present invention;

[0040] Figure 8 It is a flow chart of fault diagnosis and location of local hybrid g-good neighbor + ring diagnosis of tumor medical system based on HPMC model of the present invention;

[0041] Fig. 9 It is the time comparison of the simulation experiment results of the present invention;

[0042] Fig.10 It is a comparison diagram of the fault node diagnosis effect of the simulation experiment results of the present invention. DETAILED DESCRIPTION

[0043] The present invention is further described in detail below with reference to examples so that those skilled in the art can implement the invention with reference to the description.

[0044] It should be understood that terms such as “having”, “including” and “comprising” used herein do not exclude the existence or addition of one or more other elements or combinations thereof.

[0045] As shown in Table 1, the test rules of the HPMC model of the present invention are:

[0046] Table 1 Test rules for HPMC model

[0047] Test node u Test edge (u,v) Node v under test Test results σ(u,(u,v),v) 0 0 0 0 0 0 1 1 0 1 0 1 1 0 0 0 / 1 1 0 1 0 / 1

[0048] Figure 1 This is an example of a network structure diagram of a tumor medical system constructed under the HPMC model targeted by the present invention. In the diagram, black nodes indicate no faults, and red nodes and edge states indicate faults.

[0049] In the HPMC model, a node is used to test its neighboring nodes, and the state feedback is performed based on the test results of the nodes. The test results are represented by σ(u,(u,v),v), where u is the test node (the node on the tumor medical system), (u,v) is the test edge (the connection between two nodes in the tumor medical system, also known as the communication link), and v is the node under test (the node adjacent to the test node on the tumor medical system). The test rules are shown in Table 1. If the test result is 0, it means that the tested node and its communication link (i.e., the connection edge between the nodes) are fault-free; if the test result is 1, it means that the tested node or communication link (i.e., the connection edge between the nodes) is faulty. It is worth noting that according to the test rules of the HPMC model, each node in the tumor medical system and its communication link cannot fail at the same time, otherwise the test results fed back will be inaccurate. Since it is impossible to determine which part caused the fault, when the faulty communication link is connected to the faulty node, the accuracy of the test results will be affected. In addition, the test results are only credible if the nodes involved in the test are themselves in a fault-free state, otherwise the results will be uncertain.

[0050] However, in both the HPMC model and the PMC model, the fault diagnosis process only focuses on the global state of the system and ignores the local state of the system. For example, if system G is formed by integrating two arbitrary subsystems in some way, and the two subsystems can diagnose the maximum number of faulty processors or faulty communication links as m and n, respectively, and m is much larger than n, then the diagnostic capability of system G is limited to n, but even if the number of faulty processors and communication links is between m and n, all faulty processors or faulty communication links can be correctly identified. Therefore, only considering the global fault / no fault state will lose some local system details, which can easily lead to a decrease in the accuracy of system fault diagnosis.

[0051] In contrast, considering local fault diagnosis can more comprehensively evaluate the fault diagnosis capability of the system and greatly improve the diagnostic accuracy. At the same time, the fault diagnosis method of the PMC model is based on the situation that the neighboring nodes of a certain node may fail at the same time to obtain the maximum number of faulty nodes in the system. The possibility of this situation occurring in the actual operation of the system is low. The present invention uses the HPMC model to diagnose the local state of the system first, and the state of uncertain nodes can be judged by their neighboring nodes. According to the condition that each fault-free node has at least g good neighbors (each fault-free node has at least g fault-free neighboring nodes, g is greater than or equal to 1), the state of each uncertain node is determined in turn, thereby improving the accuracy and reliability of fault diagnosis.

[0052] However, the g-good neighbor conditional fault diagnosis strategy has certain limitations, especially in complex tumor medical systems with a ring structure. Specifically, when there are multiple nodes in the system that form a closed loop, even if a node fails and is removed, the connectivity of the entire system will not be destroyed. To address this problem, the present invention combines the concept of cycle diagnosis (ct(G)) for the first time and proposes a local mixed g-good neighbor + cycle diagnosis fault self-diagnosis method, thereby improving the diagnostic accuracy of the system in local and global states. The introduction of cycle diagnosis ensures that even if the system is divided into multiple connected subgraphs after a failure and there is a ring structure in the subgraph, all faulty nodes and communication links can still be reliably diagnosed.

[0053] By combining ring diagnosis with g-good neighbor conditions, the local mixed g-good neighbor + ring diagnosis fault diagnosis method of the present invention not only significantly improves the system's coverage of local and global faults, but also greatly improves the accuracy and efficiency of diagnosis. This method can effectively isolate faulty nodes and communication links under complex system topologies and ring structures, ensure the high reliability of tumor medical systems in complex fault environments, and provide accurate data support for diagnosis and treatment decisions. In addition, the parallelized local diagnosis strategy in the method comprehensively considers the local details of the system, greatly improves the efficiency and accuracy of fault detection, and provides a new technical path for high-reliability fault diagnosis of the system.

[0054] The present invention provides a new local hybrid fault diagnosis method under the HPMC fault diagnosis model, that is, a fault diagnosis method combining ring diagnosis and g-good neighbor conditions. This method can locate the source of the fault more conveniently, efficiently and accurately. In the tumor medical system, each key link corresponds to a node, and a sensor is set for each key link to monitor the node status in real time, detect abnormal situations in time and perform early warning processing. The specific steps are:

[0055] First, a node in the tumor medical system is randomly selected and its state is set to a fault-free state. Then, the sensors of each node will obtain corresponding values ​​(i.e., the test results σ mentioned in the present invention) based on the data collected at ordinary times. Finally, combined with the test rules of the HPMC model, the node or communication link failure is diagnosed and the sensor value is dynamically updated. Figure 2As shown, red indicates a faulty node and a faulty edge, and blue indicates that the state of the node or edge cannot be determined. Assuming that a and f are fault-free nodes, according to the test rules of the HPMC model, nodes b and e are diagnosed as faulty nodes respectively, then the states of nodes c and d, as well as the test edges between the nodes cannot be determined. Therefore, the present invention uses the local g-good neighbor condition to perform further diagnosis, starting from the neighboring nodes of the faulty node. It is observed that there is a connection between nodes c and f, and a connection between nodes d and a. According to the condition that each fault-free node has at least g (g≥1) fault-free nodes, it can be quickly diagnosed that nodes c and d are fault-free nodes. Then, from σ(c,(c,d),d)=1 and σ(d,(d,c),c)=1, it can be concluded that the test edge (c,d) is faulty.

[0056] Since the tumor medical system has redundant paths, the present invention constructs a special structure in order to better and faster diagnose the faulty nodes and communication links in the tumor medical system, and names it as the "small diamond structure", which is suitable for the diagnosis method proposed by the present invention. Figure 3 As shown in the figure. Part L in the figure fully demonstrates the fault diagnosis method based on the local mixed g-good neighbor + ring under the HPMC model. Assume that node a (such as chest CT equipment) is a fault-free node and the communication link (a, b) is fault-free. According to the test rules of the HPMC model, σ(a, (a, b), b) = 1 and σ(b, (a, b), a) = 0, then node b (diagnosis result) can be diagnosed as a faulty node. Since σ(a, (a, f), f) = 1 and σ(f, (a, f), a) = 0, the status of node f and communication link (a, f) cannot be determined according to the test rules of the HPMC model, but according to the local g-good neighbor condition, node a is connected to at least one fault-free node, so it can be determined that node f is a fault-free node and the communication link (a, f) is faulty. Since b is a faulty node, the status of nodes c and e cannot be determined. Because σ(f,(f,i),i)=0 and σ(i,(i,f),f)=0, the status of node i and communication link (f,i) can be determined to be fault-free. However, it is found that σ(i,(i,d),d)=1 and σ(d,(i,d),i)=0, so the status of node d cannot be determined, and the g-good neighbor condition cannot be used to judge the node status of c, d, and e. Obviously, in this case, it is difficult to accurately locate all fault sources by relying only on the local g-good neighbor condition.

[0057] Therefore, the present invention further introduces local hybrid cycle diagnosis to optimize the diagnosis process. The basic idea of ​​cycle diagnosis is to use the closed loop structure characteristics, that is, the characteristics of several nodes connected into a closed loop, to exclude or confirm the status of other nodes through some known fault-free paths and paths of isolated faulty nodes in the ring structure, as well as the test symptoms of the HPMC model. For example, Figure 3 The structure of the L part has nodes a, b, c, d, e, f, i connected in a closed loop. The status of nodes b and f has been obtained through the above g-good neighbor condition diagnosis, and the communication link (a, b) is fault-free, (a, f) is faulty, and the node status of c, d, e cannot be determined. The status between c, d, e and the communication link can be determined by the local mixed ring diagnosis method. In the a→b→c→d→e→f→a loop, nodes c and e are on the connection paths of b and f respectively, and node b is faulty. The local mixed ring diagnosis will check whether there are other detour paths (no fault nodes and communication links) that can connect from node a or node f to nodes c and e. Obviously, there is no detour path in the structure L that can connect from node a to c and e, so it can be inferred that c and e are also faulty nodes. Similarly, there is no path that can connect to d through node a bypassing nodes b and e, but there is a detour path f→i→d that can connect to node d. According to the characteristics of local mixed ring diagnosis, it can also be inferred that d is a fault-free node.

[0058] Example

[0059] In order to more clearly demonstrate the fault diagnosis method of local hybrid g-good neighbor + ring diagnosis based on the HPMC model, the following example of a tumor medical system is used to demonstrate the fault diagnosis process of this method in detail. In this method, the fault-free node performs fault tests on the adjacent nodes through the communication link, and combines the HPMC model test rules and g-good neighbor conditions to determine the location of the faulty node and the faulty communication link. For nodes whose status cannot be determined, the diagnosis rules of the ring diagnosis are used to diagnose the status of these nodes in turn, and then the status of the communication link (faulty or fault-free) is obtained in combination with the HPMC model test rules.

[0060] The specific scheme is as follows: In this example, a simplified tumor medical system is set up, which includes multiple inspection links and communication links interconnected with the inspection results and treatment plans. These inspection links transmit information through communication links. The goal of the present invention is to automatically detect and quickly locate the faulty nodes or faulty communication links (such as disconnected communication links or faulty nodes, etc.) in the tumor medical system through local mixed g-good neighbor conditional fault diagnosis, so as to identify the source of the fault as quickly and effectively as possible. Subsequently, all nodes in uncertain states are determined in combination with local mixed ring diagnosis, so as to comprehensively locate the source of the system's fault.

[0061] In the local hybrid g-good neighbor + ring fault diagnosis based on the HPMC model of the present invention, state 0 indicates that the node and the communication link are operating normally, and state 1 indicates that there is a fault in the node or the communication link. In addition, the present invention provides three groups of estimated parameters based on the HPMC model:

[0062] Initial state probability SP (Star Probability):

[0063] The probability that the system is in state 0 or state 1 at time t = 1;

[0064] State transition probability TP (Transition Probability):

[0065] The probability of the system transitioning from state i to state j;

[0066] Observation Probability OP (Observation Probability):

[0067] The probability of observing 0 or 1 given state i.

[0068] 1. Step 1: Initialization

[0069] First, a network structure diagram of the tumor medical system is constructed. Then, at time t=1, a local network of the tumor medical system is selected according to actual conditions or randomly, and a node is randomly selected as the initial node to determine its initial state. For example, the state of the diagnosis result a node is 0, which indicates that this node (diagnosis result a) is operating normally.

[0070] 2. Step 2: State Transfer

[0071] According to the state transition probability TP, calculate the probability of transitioning from the current state (the state of the previous step) to the next state. For example, the current state is S t (state 0 or state 1), then according to the state transition probability matrix, calculate S t Transfer to S t+1 The probability of, and select the state with a larger probability as the next state, such as Figure 4 shown.

[0072] 3. Step 3: Start from the initial node and perform breadth-first traversal algorithm

[0073] Starting from the initial node a, a breadth-first traversal is performed to traverse the degrees of each node in the tumor medical system network structure diagram. "0" indicates that there is no connection edge between node a and node b, and "1" indicates that there is a connection edge between node a and node b, thereby obtaining the node degree matrix of the tumor medical system, such as Figure 5 shown.

[0074] 4. Step 4: Fault diagnosis test based on HPMC model test rules

[0075] To test the network structure of the tumor medical system according to the test rules of the HPMC model, it is necessary to first test all nodes and test edges (communication links) in the tumor medical system network to obtain a test symptom set, and then use binary coding to build a graphical representation of the network topology. Use a relational data model to generate point-edge mixed fault sets from small to large and store them in the form of a relational data model. For each point-edge mixed fault set, perform state assignment, point-edge triple assignment, and diagnostic testing. Finally, use the test rules of the HPMC model to establish a symptom set. Specifically, Figure 6 shown.

[0076] 5. Step 5: Local mixed g-good neighbor condition fault diagnosis for tumor medical system

[0077] After obtaining the test matrix by storing the obtained symptom set in the test matrix, the depth-first traversal of the selected local network of the tumor medical system is carried out in parallel operation to find the same symptoms as those in the symptom set, and the first round of fault node and fault edge diagnosis is quickly performed and accurately located. However, for some nodes whose states cannot be determined in the "little diamond" structure, the g-good neighbor condition and local diagnosis method may not be sufficient to determine their states. The nodes whose states cannot be determined are marked as x1, x2, ..., x n .

[0078] 6. Step 6: Local Mixed Loop Fault Diagnosis of Tumor Medical System

[0079] For nodes x1, x2, ..., x n , determine the status of these nodes according to the local mixed ring diagnosis strategy. The specific diagnosis process is: ① Start searching from the neighboring nodes of the nodes whose status cannot be determined, and find multiple subgraphs containing ring structures according to the degree matrix; ② For these ring structure subgraphs containing uncertain node status, use the closed characteristics of the ring to further determine the faulty nodes (the detailed diagnosis process is the same as the diagnosis of the above-mentioned "little diamond" special structure); ③ After diagnosing the nodes whose status cannot be determined by combining the g-good neighbor and ring diagnosis methods, judge whether the status between these nodes (communication links) is faulty or not faulty according to the test symptom set of the HPMC model. The detailed steps of local mixed ring diagnosis are as follows: Figure 7 shown.

[0080] 7. Step 7: Status Update

[0081] The state probability distribution at time t=i is updated using the observation probability OP and the state transition probability TP and stored in the database.

[0082] 8. Step 8: Repeat steps 1 to 7 to locate the faulty nodes and faulty links in the entire tumor medical system.

[0083] Since the tumor medical system has too many nodes and there are unconnected parts, we can repeat steps 1 to 7 to perform local fault diagnosis on the local network part of the complete tumor medical system, determine the number of local fault nodes and communication links of the tumor medical system, and locate the fault source.

[0084] Finally, through the diagnosis method combining local hybrid g-good neighbor + ring diagnosis, the minimum number of faulty nodes and communication links in the local fault diagnosis result matrix is ​​obtained, and then according to the relationship between global and local diagnosis, the maximum number of faults in the entire tumor medical system is obtained, and all faulty nodes and communication links in the system are quickly diagnosed and located. The flow chart is as follows Figure 8 shown.

[0085] Since the diagnostic capability of the global fruit supply chain network graph is limited by the diagnostic capability of the local fruit supply chain network graph, the minimum number of fault nodes and fault communication edges of the local fruit supply chain network graph obtained by traversal is the number of fault nodes and fault communication edges of the global fruit supply chain network graph, or the maximum diagnostic capability of the global fruit supply chain system is limited by the minimum number of fault nodes and fault communication edges of the local fruit supply chain network graph. For example, system G is formed by integrating two arbitrary subsystems in some way, wherein the two subsystems can diagnose the maximum number of fault processors or fault communication links as m and n respectively, and m is much greater than n. At this time, the diagnostic capability of system G is limited by n. Therefore, even if the number of fault processors or fault communication links is between m and n, system G can only correctly identify no more than n fault processors or fault communication links. A computing device includes a processor and a memory, wherein the memory stores executable code, and when the processor executes the executable code, the fault diagnosis method combining local mixed g-good neighbor and ring diagnosis for tumor medical system described in the present invention is implemented.

[0086] The present invention can achieve the following technical effects:

[0087] 1. Based on the HPMC model, the present invention considers system-level local faults and proposes a more efficient local hybrid g-good neighbor + ring fault diagnosis method. It is not necessary to conduct full coverage testing on each link and communication link in the huge tumor medical system, but to divide the huge tumor medical system into several local network structures, select some network structures to detect the fault source, and each part can perform fault diagnosis in parallel, so that the location of the fault source can be determined more accurately and quickly, and the state of the node can be observed in real time by using sensors. The present invention sets conditions for nodes whose states cannot be determined, and limits each fault-free node to be adjacent to at least one fault-free node, which improves the diagnosis rate of fault diagnosis. In addition, for nodes whose states cannot be determined under local g-good neighbor conditions, local hybrid ring diagnosis is used to achieve rapid diagnosis and locate the fault source. The HPMC model is based on a graph theory model, which regards the processor as a node in the graph and the communication link as an edge in the graph. The fault-free node issues a test to its neighbor node, and determines the state of the tested node and the test edge according to the test rules.

[0088] 2. Based on the HPMC fault diagnosis model, the present invention first determines the neighbor nodes of each node through the breadth-first algorithm, and then uses the depth-first algorithm combined with the test rules of the HPMC model to determine the location of the local faulty nodes and the faulty communication links through the test results fed back each time. Finally, through the relationship between the local and the global, the faulty nodes and communication links of the entire tumor medical system are quickly located.

[0089] 3. The tumor medical system based on the graph theory model of the present invention is to abstract the entire tumor medical system into a graph, with nodes and edges representing various examinations and data in the tumor medical system, and data and results as communication links, and describe their operating status through state attributes. The local mixed g-good neighbor + ring fault diagnosis optimization of the tumor medical system is performed through the test rules of the HPMC model and the constructed "small diamond" structure. In the diagnosis process, for nodes whose local g-good neighbor conditions cannot be determined, local mixed ring diagnosis is used to further determine the state of the node, so as to achieve more comprehensive fault identification.

[0090] Simulation experiment

[0091] A simulation experiment of the present invention is implemented in a python compilation environment, wherein the hardware environment includes a Windows 10 operating system, an Intel Core i7-9750H@2.60GHz processor, and an NVIDIA RTX3060 graphics card. The software environment uses a Python 3.8 compiler, and the integrated development environment is an IDE. The specific steps are as follows:

[0092] Step (1): Construct a network topology diagram of the tumor medical system. Assuming that the tumor medical system has 256 nodes and 1024 edges, construct a network structure diagram based on the properties of the tumor medical system.

[0093] Step (2): Divide the tumor medical system into several local network structure diagrams. Use a uniform division method to ensure that the size of each local network is moderate, which is convenient for analysis and diagnosis. At the same time, the connection relationship between the nodes is fully considered during the division to ensure that the nodes within each local network have a high degree of connectivity, so that local information can be fully utilized during fault diagnosis. This experiment is divided into four local network structure diagrams.

[0094] Step (3): Use the depth-first traversal algorithm to obtain a symptom set, use the test matrix to store the obtained symptom set to obtain the test matrix, perform local fault diagnosis according to the HPMC model test rules and the g-good neighbor condition, and store the nodes whose states cannot be determined in the matrix NS.

[0095] Step (4): Combine the local hybrid ring diagnosis and the degree matrix to determine the state of the nodes in the matrix NS, and then combine the test rules of the HPMC model to quickly diagnose the faulty communication link and achieve comprehensive and accurate fault diagnosis.

[0096] Step (5): Obtain the total number of local faulty nodes and faulty edges, and based on the relationship between the local and global (the total number of global faulty nodes and faulty edges is the minimum number of local faulty nodes and faulty edges), obtain the number of faulty nodes and faulty communication edges of the global tumor medical system.

[0097] In order to verify the effectiveness of the local hybrid g-good neighbor + ring fault diagnosis method of the present invention, a simulation experiment based on the HPMC model comprehensively evaluates the performance of different diagnosis strategies in a complex network environment. The experiment compares the performance of three diagnosis methods:

[0098] 1. Local diagnosis strategy based on HPMC model (compared with the present invention, g-good neighbor + ring diagnosis strategy is not used);

[0099] This strategy only relies on the local information of nodes in the network for fault diagnosis, and does not introduce the g-good neighbor + ring diagnosis strategy. It analyzes based on the state of the node itself, and there is a problem of reduced diagnostic accuracy due to insufficient information. Specifically, Fig. 9 and Fig.10 According to the experimental results, the diagnosis time is 0.43 ms, 0.31 ms, 0.36 ms, and 0.38 ms, and the number of faulty nodes diagnosed is 3, 6, 2, and 4. Since only local information is considered, the accurate location of the fault is limited by the local information.

[0100] 2. A local hybrid diagnosis strategy based on the HPMC model (compared with the present invention, a “point-edge hybrid diagnosis” method is used, and the g-good neighbor + ring diagnosis strategy is not used);

[0101] Based on the local diagnosis of the HPMC model, this strategy introduces the "point-edge hybrid diagnosis" method, which improves the accuracy of fault diagnosis by analyzing local nodes and their connection relationships. Compared with simple local diagnosis, the hybrid strategy has improved the diagnosis time and the number of faulty nodes. Specifically, the diagnosis time is 0.43 milliseconds, 0.25 milliseconds, 0.32 milliseconds, and 0.36 milliseconds, respectively, and the number of faulty nodes diagnosed is 6, 7, 7, and 6. However, this strategy still faces certain computational complexity and possible diagnostic blind spots, which may affect the diagnosis time and accuracy, especially when dealing with large or highly dynamic networks.

[0102] 3. And the local mixed g-good neighbor + ring diagnostic strategy based on the HPMC model proposed by the present invention.

[0103] The local hybrid g-good neighbor + ring diagnosis strategy based on the HPMC model proposed in this paper combines g-good neighbor node identification and ring structure diagnosis, and can comprehensively improve the accuracy and precision of fault diagnosis based on local information. By identifying the adjacency relationship between local loops and nodes, this strategy effectively overcomes the shortcomings of traditional local diagnosis strategies, especially in complex network environments, it can accurately identify various types of faults.

[0104] Experimental results ( Fig. 9 , Fig.10 ) shows that the strategy is superior to the other two methods in terms of fault diagnosis time and accuracy. The diagnosis time is 0.156 ms, 0.09 ms, 0.20 ms, and 0.23 ms, respectively, and the number of fault nodes is 9, 7, 8, and 6, demonstrating the significant advantages of the strategy in terms of efficiency and accuracy.

[0105] like Fig. 9 and Fig.10 As shown, from the time comparison results and fault diagnosis results of the simulation experiment, it can be seen that the local mixed g-good neighbor + ring fault diagnosis method proposed in the present invention is superior to other local diagnosis methods in terms of diagnostic accuracy, robustness and efficiency. The method of the present invention enhances the fault adaptive ability of the system under complex network structures by introducing local mixed ring diagnosis technology, and significantly improves the accuracy and coverage of diagnosis. Especially in tumor medical systems, the system scale is large, the topology is complex, and the failure modes are diverse. The method of the present invention can quickly and accurately identify faulty processors and communication links, ensure the high reliability and connectivity of the system in complex fault environments, and provide accurate support for medical decision-making.

[0106] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation methods. It can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. In addition, the present invention belongs to the field of network system diagnosis technology, specifically to system diagnosis models and network system diagnosis degrees, etc. In addition, the fault diagnosis method / system combining local mixed g-good neighbor and ring diagnosis for tumor medical systems based on the HPMC model is a preferred embodiment of the present invention, which is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A fault diagnosis method combining local hybrid g-good neighbor and ring diagnosis for tumor medical system, characterized in that: include: S1. Starting from a node in a non-faulty state in the tumor medical system, according to the HPMC model test rules, diagnose the faults of adjacent nodes and communication links, and mark the nodes and communication links whose states cannot be determined; S2, using the local g-good neighbor condition to further diagnose the failure of nodes and communication links whose states cannot be determined, and marking the remaining nodes and communication links whose states cannot be determined; S3. Start searching the ring structure from the adjacent nodes of the remaining nodes whose states cannot be determined, use the ring diagnosis rules to further diagnose the faults of the remaining nodes, and combine the HPMC model test rules to diagnose the faults of the communication links between the remaining nodes.

2. The fault diagnosis method combining local hybrid g-good neighbor and ring diagnosis for tumor medical system according to claim 1, characterized in that: The local g-good neighbor condition is that there are at least g fault-free nodes for each fault-free node, where g is greater than or equal to 1.

3. The fault diagnosis method combining local hybrid g-good neighbor and ring diagnosis for tumor medical system according to claim 2, characterized in that: The ring structure includes: node a, node b, node c, node d, node e and node f which are sequentially connected to form a closed loop via communication links.

4. The fault diagnosis method combining local hybrid g-good neighbor and ring diagnosis for tumor medical system according to claim 1 or 3, characterized in that: First, the tumor medical system is divided into several local network structures, and the local network structure is selected to execute steps S1 to S3 to obtain the minimum number of fault nodes and communication links in the fault diagnosis results of all local network structures. Then, based on the relationship between global and local diagnosis, the maximum number of faults in the entire tumor medical system is obtained.

5. The fault diagnosis method combining local hybrid g-good neighbor and ring diagnosis for tumor medical system according to claim 4, characterized in that: First, a breadth-first traversal of the tumor medical system is performed to generate a symptom set, and then a depth-first traversal of the local network structure is performed based on the symptom set to diagnose node and communication link failures.

6. The fault diagnosis method combining local hybrid g-good neighbor and ring diagnosis for tumor medical system according to claim 1, characterized in that: Specifically include: Step 1: Construct the network structure of the tumor medical system and divide it into several local network structures, select the initial nodes, and set the initial state; Step 2: Starting from the initial node, perform breadth-first traversal to traverse the degree of each node in the tumor medical system network structure to obtain the node degree matrix of the tumor medical system; Step 3: Test the network structure of the tumor medical system according to the test rules of the HPMC model, obtain the test symptom set, perform binary encoding, generate a point-edge mixed fault set, and establish a symptom set through state assignment, point-edge triple assignment and diagnostic testing; Step 4: Perform a depth-first traversal of the local network structure of the selected tumor medical system to find the same symptoms as those in the symptom set, and quickly perform the first round of fault node and fault edge diagnosis and location, and mark the nodes whose status cannot be determined; Step 5: For nodes whose states cannot be determined, start searching from the neighboring nodes of the nodes whose states cannot be determined, and find multiple ring structures respectively according to the degree matrix; for these ring structures containing uncertain node states, further determine the faulty nodes by using the closed characteristics of the rings; after completing the diagnosis of the nodes whose states cannot be determined, determine whether the states of the communication links between these nodes are faulty or not according to the test symptom set of the HPMC model; Step 6: Update status; Step 7: Repeat steps 1 to 6 to locate the faulty nodes and faulty links in the entire tumor medical system.

7. A computing device comprising a processor and a memory, characterized in that: The memory stores executable codes, and when the processor executes the executable codes, the method according to any one of claims 1 to 6 is implemented.

Images