Online isolation operation and maintenance management platform and method based on modular low-temperature economizer

Abstract

The application discloses an online isolation operation and maintenance management platform and method based on a modular low-temperature economizer, and particularly relates to the field of low-temperature economizer online maintenance control, and comprises the following steps: a cloud end constructs an isolation graph according to a target heat exchange section number in a maintenance work order and the connection relationship of each heat exchange section, water inlet branch, water return branch and flue gas damper, and generates a process chain according to the chronological relationship of flow interruption, pressure loss, liquid discharge and temperature drop; when the remaining heat exchange sections remain in operation and the total water inlet flow remains unchanged, the edge end applies the same amplitude reverse valve position adjustment to the water inlet branch pointed to by the maintenance work order and the water inlet branch with a water inlet path to the target heat exchange section in the isolation graph one by one; the application realizes the correlation verification of the maintenance module and the actual isolation component and the continuous determination of the isolation state by means of isolation graph construction, branch response matching, isolation state probability recursion and process closed-loop control, so as to solve the problem that the target module, the isolation component and the field response object are misaligned in the existing online maintenance process.

Description

Technical Field

[0001] This invention relates to the field of online maintenance and control technology for cryogenic economizers, and more specifically, to an online isolation operation and maintenance management platform and method based on modular cryogenic economizers. Background Technology

[0002] Online maintenance of modular cryogenic economizers is usually based on the results of operation monitoring to determine the module to be handled. Then, the operation and maintenance personnel close the corresponding inlet valve, return valve and flue gas side isolation component according to the work order. After the pressure, flow and temperature meet the maintenance requirements, pressure relief, drainage and disassembly operations are carried out.

[0003] When multiple modules operate in parallel and share inlet and outlet water headers, flue gas channels, and centralized valve groups, the module number, field valve group number, control point, and measurement point in the work order must be accurately matched. Simultaneously, the isolation states of flow loss, pressure loss, and temperature drop must remain valid throughout the maintenance period. Otherwise, even if valves and baffles operate as instructed, adjacent modules may be actually isolated due to misalignment between module and valve group mappings, or the confirmed isolation conditions may fail due to valve leakage, media backflow, or a return to normal status. This can manifest on-site as the target module's flow not returning to zero, abnormal temperature decay in non-target modules, pressure re-establishment after depressurization, or media inflow into the corresponding module even after maintenance procedures have been approved. Current maintenance methods typically handle isolation object verification and maintenance procedure confirmation separately, making it difficult to continuously verify the actual isolated object based on on-site responses and to simultaneously revoke subsequent work permissions when the isolation status changes.

[0004] Therefore, the technical problem to be solved in this application is how to ensure that the module specified in the work order, the actual isolated component and the response object on site are consistent during the non-stop maintenance of the modular cryogenic economizer, and how to control the step-by-step release and relocking of the maintenance process based on the continuous change of the isolation status. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an online isolation operation and maintenance management platform and method based on a modular cryogenic economizer. Through isolation diagram construction, branch response matching, isolation state probability recursion, and process closed-loop control, the platform achieves the correlation verification between the maintenance module and the actual isolated components, as well as the continuous determination of the isolation state, thereby solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an online isolation operation and maintenance management method based on a modular cryogenic economizer, comprising:

[0007] S1. The cloud platform constructs an isolation diagram based on the target heat exchange section number in the maintenance work order, as well as the connection relationship between each heat exchange section, inlet branch, return branch and flue gas damper, and generates a process chain according to the sequence of flow interruption, pressure loss, liquid drainage and cooling.

[0008] S2. When the edge end continues to operate in the other heat exchange sections and the total inlet flow remains unchanged, apply the same amplitude reverse valve position adjustment to the inlet branch pointed to by the maintenance work order and the inlet branch with an inlet path to the target heat exchange section in the isolation diagram one by one. Generate a response sequence according to the change sign, start position and recovery position of the outlet temperature of each heat exchange section and upload it to the cloud.

[0009] S3. The cloud uses the change symbol, start position and recovery position in the response sequence as matching features, performs secondary assignment under the constraint of the isolation diagram, and generates a branch and heat exchange section mapping matrix through projection iteration. The inlet branch with the target heat exchange section sorting number 1 and its connected return branch and flue gas damper are used as isolation objects, and the mapping value between the target heat exchange section and the inlet branch is output.

[0010] S4. At the edge, the system performs flue gas isolation, inlet and outlet water cut-off, pressure relief, and liquid drainage on the isolated object, and collects the flow rate, pressure, and temperature difference of the target heat exchange section to generate a state sequence. At the cloud, the system performs Bayesian recursion based on the mapping value as a priori, and outputs the probability of success and failure of flow interruption, pressure loss, and temperature drop.

[0011] S5. Based on the mapping value and the probability of flow interruption, pressure loss and temperature drop, the cloud issues maintenance release instructions according to the process chain. When the flow interruption, pressure loss or temperature drop status fails, the subsequent maintenance process is locked along the process chain from the node associated with the failure status.

[0012] In a preferred embodiment, S1 includes:

[0013] S1-1, The cloud uses the heat exchange section number, water inlet branch number, water return branch number and flue gas damper number as point identifiers, and the water flow direction or flue gas flow direction between two points as the edge direction. It searches backward from the target heat exchange section to the water inlet header, water return header and flue gas inlet respectively, and generates the water inlet path, water return path and flue gas path composed of point identifiers and the connection order of adjacent points.

[0014] S1-2. Form isolation groups by combining the inlet branches in the inlet path, the return branches in the return path, and the flue gas baffles in the flue gas path. Remove the points and connected edges in the isolation groups one by one, and determine whether there are still directed paths from the inlet header, return header, and flue gas inlet to the target heat exchange section. Keep the isolation groups where all three types of directed paths are cut off. When the number of points in the isolation groups is the same, sort them in ascending order according to the total number of edges from each point to the target heat exchange section, and take the isolation group with the first position in the sort to generate the isolation diagram.

[0015] S1-3. The process is directed by the closure of the flue gas damper to the closure of the inlet branch and the return branch, and the closure of the inlet branch and the return branch together to the depressurization. The depressurization leads to the liquid discharge, and the liquid discharge leads to the cooling. A directed edge for the process is established. In each round, the process with zero incoming edges is selected to generate the current process position. The outgoing edges of the current process position are deleted and the selection continues until a process chain arranged in the order of execution is generated.

[0016] In a preferred embodiment, S2 includes:

[0017] S2-1. At the edge end, the sum of the flow rates of each operating branch before adjustment is used as the baseline total. First, the valve position of the branch pointed to by the work order is increased by one control resolution unit. Then, the valve positions are decreased sequentially according to the flow rate ratio of the remaining operating branches before adjustment, until the sum of the flow rates of each operating branch is restored to the baseline total, generating a positive adjustment record. After restoring each operating branch to the valve position before adjustment, the valve position of the branch pointed to by the work order is decreased by one control resolution unit. Then, the valve positions are increased sequentially according to the flow rate ratio of the remaining operating branches before adjustment, until the sum of the flow rates of each operating branch is restored to the baseline total, generating a reverse adjustment record.

[0018] S2-2. Using the outlet temperature of each heat exchange section at the start of the forward adjustment as the reference temperature, calculate the difference between the outlet temperature and the reference temperature according to a uniform sampling interval. The first sampling position that satisfies the same sign for two adjacent temperature differences and the absolute value of the latter is greater than the absolute value of the former is taken as the starting position, and the sign of the temperature difference at the starting position is taken as the change sign. From the start of the reverse adjustment, compare the absolute values ​​of the temperature difference according to the sampling order, and take the sampling position before the absolute value of the temperature difference changes from decreasing to increasing as the recovery position.

[0019] S2-3. Arrange the forward adjustment records, reverse adjustment records, change symbols, start position and recovery position according to the heat exchange section number, generate the response sequence and upload it to the cloud.

[0020] In a preferred embodiment, S3 includes:

[0021] S3-1. The cloud takes the change symbol, start change position and recovery position in the response sequence as input, calculates the number of inconsistent change symbols, the difference of start change position sequence number and the difference of recovery position sequence number between the water inlet branch and the heat exchange section as sign difference, start change difference and recovery difference respectively, and uses the sum of sign difference, start change difference and recovery difference as matching cost.

[0022] S3-2. Based on the water inlet path in the isolation diagram, remove the combination of water inlet branch and heat exchange section that does not have a water inlet path. With each water inlet branch associated with one heat exchange section and each heat exchange section associated with one water inlet branch as a constraint, perform secondary assignment calculation on the matching cost of the remaining combinations to generate an initial mapping matrix. Update the non-zero elements in the initial mapping matrix according to the decreasing direction of the matching cost, and normalize the updated matrix elements by dividing them by the sum of the elements in their row and the sum of the elements in their column, respectively, to generate a mapping matrix.

[0023] In a preferred embodiment, S3 further includes:

[0024] S3-3. Read the column containing the target heat exchange section in the mapping matrix, sort the mapping values ​​in the column in descending order, determine the inlet branch corresponding to the mapping value with the sorting number 1 as the candidate inlet branch, and read the return water branch and flue gas damper connected to the candidate inlet branch along the isolation diagram; determine whether there is an inlet path between the candidate inlet branch and the target heat exchange section, whether the sorting number of the mapping value is 1, and whether the number of consistent change signs in the response sequence is equal to the number of response points. When all three judgment results are true, the candidate inlet branch is determined as the target inlet branch; otherwise, a mapping conflict result is generated and the response sequence is regenerated at the edge end.

[0025] S3-4. The return water branch and flue gas damper connected to the target water inlet branch are combined with the target water inlet branch to form an isolation object, and the matrix element value at the intersection of the column where the target heat exchange section is located and the row where the target water inlet branch is located is determined as the mapping value; according to the isolation diagram, it is verified whether the water inlet branch, return water branch and flue gas damper in the isolation object are in the same target heat exchange section isolation path. When the three are in the same isolation path, the isolation object and the mapping value are sent to the edge end.

[0026] In a preferred embodiment, S4 includes:

[0027] S4-1. The edge end outputs a shutdown command to the flue gas damper, the inlet branch valve, and the return branch valve in sequence according to the flue gas damper number, the inlet branch valve number, and the return branch valve number in the isolated object. It also collects the inlet flow rate, internal pressure, and inlet-outlet temperature difference of the target heat exchange section and generates the flow rate sequence, pressure sequence, and temperature difference sequence according to the sampling time.

[0028] S4-2. Subtract the (k-1)th sample value from the kth sample value in the flow rate sequence, pressure sequence, and temperature difference sequence to obtain the flow rate change, pressure change, and temperature difference change, respectively. Generate the flow rate change bit, pressure change bit, and temperature difference change bit according to the signs of the flow rate change, pressure change, and temperature difference change. Subtract the number of rising bits from the number of falling bits as the flow interruption factor, pressure loss factor, and temperature drop factor, respectively.

[0029] In a preferred embodiment, S4 further includes:

[0030] S4-3. Using the mapping value as the prior to the establishment and the mapping value minus one as the prior to the destruction, the number of falling positions in the flow interruption factor, pressure loss factor, and cooling factor are used as the evidence of establishment and the number of rising positions are used as the evidence of destruction. Bayesian recursion is performed on flow interruption, pressure loss, and cooling in sequence to generate the probability of flow interruption and the probability of flow interruption destruction, the probability of pressure loss and the probability of pressure loss destruction, and the probability of cooling and the probability of cooling destruction.

[0031] S4-4. Read the success probability and failure probability of each according to the order of flow interruption, pressure loss and cooling. When the success probability of flow interruption is higher than the failure probability of flow interruption, the success probability of pressure loss is higher than the failure probability of pressure loss and the success probability of cooling is higher than the failure probability of cooling, upload the isolation state sequence to the cloud. Otherwise, determine the first sampling bit whose success probability is not higher than the failure probability as the isolation change bit and upload it to the cloud.

[0032] In a preferred embodiment, S5 includes:

[0033] S5-1: Read the mapping value from the cloud and read the probability of the current interruption state, the pressure loss state, and the cooling state at the current sampling point; combine the mapping value, the probability of success, and the probability of failure to form the object value, the state value, and the failure value, respectively.

[0034] S5-2. Based on the directed edge relationships between flue gas isolation, water inlet / outlet cut-off, pressure relief, and liquid discharge in the process chain, the release state of the flue gas isolation node is used as the prerequisite state of the water inlet / outlet cut-off node, the release state of the water inlet / outlet cut-off node is used as the prerequisite state of the pressure relief node, and the release state of the pressure relief node is used as the prerequisite state of the liquid discharge node, thus generating a process dependency sequence.

[0035] In a preferred embodiment, S5 further includes:

[0036] S5-3. Using the following as the inputs for release calculation: the probability of the mapping value belonging to the inlet branch with the corresponding sorting number 1 of the target heat exchange section, the probability of the flow interruption being higher than the probability of the flow interruption failure, the probability of the pressure loss being higher than the probability of the pressure loss failure, and the probability of the cooling being higher than the probability of the cooling failure; the results of each state being established are executed and calculated according to the process dependency sequence; when the state result corresponding to the current process is established and the release state of the previous process is 1, the release state of the current process is set to 1 and a maintenance release command is issued; otherwise, the release state of the current process is set to 0.

[0037] S5-4. Read the first process node with a release status of 0, and traverse each node along the outgoing edge of the process chain. Set the release status of the traversed process node to 0 and generate a locking command. When the corresponding status of the first abnormal process node in the subsequent sampling bits is restored to true, check the release status of the preceding process node along the incoming edge of the process chain. Only unlock when all preceding processes have a release status of 1.

[0038] In a preferred embodiment, the online isolation operation and maintenance management platform based on the modular cryogenic economizer includes:

[0039] The isolation modeling module is used to construct an isolation diagram in the cloud based on the target heat exchange section number in the maintenance work order, as well as the connection relationship of each heat exchange section, inlet branch, return branch and flue gas damper, and generate a process chain according to the sequence of flow interruption, pressure loss, liquid discharge and cooling.

[0040] The response generation module is used to apply the same amplitude reverse valve position adjustment to the inlet branch pointed to by the maintenance work order and the inlet branch with an inlet path to the target heat exchange section in the isolation diagram when the edge end keeps running in the other heat exchange sections and the total inlet flow remains unchanged. It generates a response sequence according to the change sign, start position and recovery position of the outlet temperature of each heat exchange section and uploads it to the cloud.

[0041] The object matching module is used in the cloud to perform secondary assignment under the constraint of the isolation graph by using the change symbol, start position and recovery position in the response sequence as matching features. It generates a branch and heat exchange section mapping matrix through projection iteration, and uses the inlet branch with the target heat exchange section sorting number 1 and its connected return branch and flue gas damper as isolation objects, and outputs the mapping value between the target heat exchange section and the inlet branch.

[0042] The state determination module is used at the edge to perform flue gas isolation, inlet and outlet water cut-off, pressure relief and liquid drainage on the isolated object, and collect the flow rate, pressure and temperature difference of the target heat exchange section to generate a state sequence; the cloud performs Bayesian change point recursion with the mapping value as a priori, and outputs the probability of success and failure of flow interruption, pressure loss and temperature drop;

[0043] The process control module is used to issue maintenance release instructions in the cloud according to the mapping value and the probability of flow interruption, pressure loss and temperature drop. When the flow interruption, pressure loss or temperature drop status fails, the subsequent maintenance process is locked along the process chain from the node associated with the failure status.

[0044] The technical effects and advantages of this invention are as follows:

[0045] 1. First, the range of the inlet water branch, return water branch and flue gas damper of the target heat exchange section is defined by the isolation diagram. Then, the temperature change symbol, start position and recovery position generated by the forward and reverse valve position adjustment of each candidate inlet water branch are used to verify the actual response object. The mapping result is continuously associated with the flow interruption, pressure loss and cooling status. This can reduce the inconsistency between the heat exchange section specified in the work order, the isolation component on site and the actual affected heat exchange section during non-stop maintenance, as well as the situation where maintenance procedures continue to be released after the isolation status fails.

[0046] 2. By deleting the isolation combination nodes and then re-searching the directed paths of water inlet, water outlet, and flue gas, only the isolation combination where all three paths are cut off is retained, which can eliminate isolation schemes that still have water-side or flue gas-side connections.

[0047] 3. By combining the difference in sign of change, the difference in starting position, and the difference in recovery position into a matching cost, and by assigning branches and heat exchange sections under the constraints of the inlet water path, it is possible to exclude combinations that only have similar temperature changes but do not have actual inlet water connections.

[0048] 4. By recursively estimating the probability of occurrence and failure of flow, pressure loss and temperature drop based on the falling and rising positions of adjacent sampled values ​​of flow, pressure and temperature difference, it is possible to identify the recovery of isolation state caused by valve internal leakage, medium backflow or pressure re-establishment.

[0049] 5. By writing the isolation status result into the process dependency sequence, releasing the subsequent process after the preceding process is released, and locking zero-sum from the first abnormal node to the next, it is possible to prevent pressure relief, drainage and subsequent maintenance from continuing to be executed when the isolation conditions are not met. Attached Figure Description

[0050] Figure 1 This is a roadmap for the online isolation operation and maintenance management technology of the present invention.

[0051] Figure 2 This is a diagram of the online isolation operation and maintenance management platform of the present invention. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] Refer to the instruction manual appendix Figure 1-2 The present invention provides an online isolation operation and maintenance management method based on a modular cryogenic economizer, comprising:

[0054] S1. The cloud platform constructs an isolation diagram based on the target heat exchange section number in the maintenance work order, as well as the connection relationship between each heat exchange section, inlet branch, return branch and flue gas damper, and generates a process chain according to the sequence of flow interruption, pressure loss, liquid drainage and cooling.

[0055] In this embodiment, in order to determine the isolation target during the non-stop maintenance of the modular cryogenic economizer, it is necessary to first construct the isolation topology of the target heat exchange section based on the actual connection relationship of the equipment. Since the cryogenic economizer is usually composed of multiple parallel heat exchange sections, each heat exchange section is connected to an independent inlet water branch, return water branch and flue gas regulating component. When carrying out maintenance under non-stop conditions, it is necessary to avoid selecting isolation points based solely on equipment number or manual experience, which may lead to the mis-isolation of non-target heat exchange sections.

[0056] Therefore, this implementation converts the equipment connection relationship into an isolation graph containing nodes and directional edges, reverse-tracks the inlet water path, return water path, and flue gas path corresponding to the target heat exchange section, and filters isolation combinations based on path cutting relationships. Simultaneously, it generates subsequent execution process chains based on the dependencies between isolation actions, establishing a correlation between isolation object selection and the order of maintenance actions. This implementation process includes the following steps:

[0057] S1-1. Obtain the topology data of the low-temperature economizer equipment from the cloud. The equipment topology data includes the heat exchange section number, inlet branch number, return branch number, flue gas damper number, inlet main pipe number, return main pipe number, and flue gas inlet number.

[0058] Each connection record includes the start point number, end point number, medium type, and flow direction; the cloud uses the heat exchange section number, inlet branch number, return branch number, and flue gas damper number as point identifiers to convert the connection relationship between adjacent devices into the edge relationship in the graph;

[0059] Specifically, for the water-side path, the cloud first reads the target heat exchange section number and searches along the reverse edge of the water inlet direction from the corresponding node of the target heat exchange section, level by level, to find the water inlet branch and water inlet header connected to the target heat exchange section, forming a node sequence from the target heat exchange section node to the water inlet header node; at the same time, it searches along the return water flow direction from the target heat exchange section node to find the return water branch and return water header connected to it, forming the return water path; for the flue gas-side path, the cloud searches along the opposite direction of the flue gas flow from the target heat exchange section node to the flue gas inlet node according to the flue gas flow direction, to obtain the flue gas path; during the path generation process, each time an adjacent device node is searched, the current node number and the connection direction between the current node and the next node are recorded, forming path data of "node number + adjacent node sequence";

[0060] For example, if the target heat exchange section is numbered H1, the connected inlet branch is numbered J1, and the inlet header is numbered M1, then the inlet path is recorded as M1→J1→H1; the return path is recorded as H1→R1→RM1; and the flue gas path is recorded as G1→D1→H1, where D1 represents the flue gas damper number. In this way, the equipment connection relationship is converted into directed graph data that can be used for path analysis.

[0061] S1-2. The cloud platform establishes candidate isolation combinations based on the generated inlet water path, return water path, and flue gas path. Specifically, the cloud platform extracts the inlet water branch directly connected to the target heat exchange section from the inlet water path, extracts the return water branch connected to the target heat exchange section from the return water path, and extracts the flue gas damper controlling the flue gas flow of the target heat exchange section from the flue gas path. The cloud platform then combines one inlet water branch, one return water branch, and one flue gas damper into a candidate isolation combination.

[0062] For each candidate isolation combination, the cloud copies the current isolation graph as a path verification graph, and deletes the inlet branch node, return branch node, and flue gas damper node contained in the candidate isolation combination from the path verification graph, while also deleting all edges connected to the above nodes; after deletion, the cloud performs three types of path search respectively:

[0063] The first type of path search starts from the inlet header node and searches for the target heat exchange section node along the water flow direction; if the target heat exchange section can still be found, it means that the candidate isolation combination cannot cut off the inlet path.

[0064] The second type of path search starts from the return water header node and searches for the target heat exchange section node along the return water path direction; if the target heat exchange section can still be found, it means that the candidate isolation combination cannot cut off the return water path.

[0065] The third type of path search starts from the flue gas inlet node and searches for the target heat exchange section node along the flue gas flow direction; if the target heat exchange section can still be found, it means that the candidate isolation combination cannot cut off the flue gas path; the candidate isolation combination is retained only when the target heat exchange section cannot be reached by any of the three path searches; when multiple candidate isolation combinations meet the cutting-off condition, the cloud further calculates the path distance of each candidate isolation combination.

[0066] The path distance is the sum of the number of edges between each node in the candidate isolation combination and the target heat exchange section node. For example, if a candidate combination includes inlet branch node J1, return branch node R1, and flue gas damper node D1, then the number of edges from J1 to H1, R1 to H1, and D1 to H1 are calculated respectively, and the three numbers are added together to obtain the path distance of the candidate isolation combination. The cloud sorts the candidate isolation combinations in ascending order of path distance and selects the candidate isolation combination with the first position in the sort as the target isolation path in the isolation graph. S1-3, The cloud establishes directed edges of the process according to the isolation action relationship in the isolation graph.

[0067] Specifically, the cloud takes the closure of the flue gas damper as the initial isolation action, and the closure actions of the inlet branch and the return branch, which can only be executed after the closure of the flue gas damper, are respectively taken as subsequent actions. Therefore, directed edges are established between the flue gas damper closure node and the inlet branch closure node and the return branch closure node, respectively. Since the depressurization process needs to be based on the completion of the closure of the inlet branch and the return branch, the cloud reads the inlet branch closure node and the return branch closure node and points both nodes to the depressurization node.

[0068] The draining action needs to be performed after the pressure inside the target heat exchange section decreases, so a directed edge is established from the pressure loss node to the draining node; the cooling process needs to be performed after the draining is completed, so a directed edge is established from the draining node to the cooling node; after the directed edges are established, the cloud calculates the number of incoming edges for each process node; during the initial calculation, the process node with 0 incoming edges is taken as the current executable process position; for example, the flue gas damper closing node has no preceding node, so its number of incoming edges is 0, and it is determined as the first process position; after determining the current process position, the cloud deletes the output edge of the node corresponding to the process position, and recalculates the number of incoming edges for the remaining nodes, and continues to select a new node with 0 incoming edges; for the case where the inlet branch closing node and the return branch closing node both satisfy the condition of having 0 incoming edges, the two nodes are assigned to the same process position, indicating that the next process can only proceed after both actions are completed; by deleting the output edges of the determined process positions round by round until all process nodes are sorted, a process chain arranged according to the execution order is generated;

[0069] Through the above isolation diagram construction process, the cloud can select isolation combinations that meet the requirements of bidirectional isolation of flue gas and water flow based on the actual connection relationship of the target heat exchange section, and use the directional relationship of the process to constrain the execution order of isolation actions, so as to avoid the problem that the medium is still connected in non-target areas due to isolation judgment based on a single valve status.

[0070] In practical applications: For example, a low-temperature economizer contains 6 parallel heat exchange sections, with heat exchange section H3 being the target for maintenance. After reading the equipment topology data, the cloud platform finds that H3 is connected to the inlet water branch J3, the return water branch R3, and the flue gas damper D3. It also discovers that there are other branches besides J3, R3, and D3 that can bypass H3 to form a medium passage. The cloud platform deletes isolation nodes from different candidate combinations to perform path searches, and finally selects the combination that cannot reach H3 from the inlet water header, the return water header, and the flue gas inlet after deleting J3, R3, and D3. This combination is then selected as the target isolation combination. Subsequently, the cloud platform generates a process chain based on the dependencies between flue gas damper closure, inlet and return water cut-off, pressure loss, liquid drainage, and cooling, enabling the edge end to perform online isolation operations according to the generated process sequence.

[0071] S2. When the edge end continues to operate in the other heat exchange sections and the total inlet flow remains unchanged, apply the same amplitude reverse valve position adjustment to the inlet branch pointed to by the maintenance work order and the inlet branch with an inlet path to the target heat exchange section in the isolation diagram one by one. Generate a response sequence according to the change sign, start position and recovery position of the outlet temperature of each heat exchange section and upload it to the cloud.

[0072] In this embodiment, in order to determine the actual response relationship between the target heat exchange section and the inlet branch, the edge end does not directly determine the heat exchange section ownership based on the valve number or topological relationship. Instead, while the other heat exchange sections are in operation, a controlled disturbance is applied to each candidate inlet branch, and the response characteristics are extracted by using the change in outlet temperature before and after the disturbance.

[0073] By creating flow disturbances in opposite directions through positive and negative adjustments, the heat exchange sections affected by the same inlet branch are caused to exhibit directional temperature changes. Response characteristics are then generated based on the location and recovery process of these temperature changes, which are used for matching calculations between subsequent branches and heat exchange sections. This implementation process includes the following steps:

[0074] S2-1. The edge end first reads the real-time flow rate of each operating branch before adjustment and calculates the sum of the flow rates of each operating branch as the baseline total. The operating branches before adjustment include the inlet branches corresponding to the heat exchange section that are not currently involved in valve position adjustment, as well as the inlet branches to be adjusted. Specifically, the edge end collects the flow rate value of each inlet branch before valve position adjustment, adds up all the flow rate values ​​to obtain the baseline total, which is used to determine whether the compensation adjustment is completed during the valve position disturbance process.

[0075] After completing the baseline total calculation, the inlet branch pointed to by the current work order is selected as the current adjustment branch at the edge end, and the valve position code of the current adjustment branch is increased by one control resolution unit; the control resolution unit is taken from the difference between adjacent valve position codes that the valve controller can recognize. For example, when the valve controller uses integer code control, the code value is increased by 1 as a valve position adjustment unit.

[0076] After the valve position is increased, the edge end reads the flow ratio of the remaining operating branches before adjustment. The flow ratio is the ratio of the flow value of a single operating branch before adjustment to the baseline total. The remaining operating branches are arranged from high to low according to the flow ratio. The edge end decreases the valve position of the remaining operating branches in the order of arrangement, decreasing by one control resolution unit each time. After each adjustment, the flow of each operating branch is re-acquired and the total flow is calculated.

[0077] When the total flow rate has not recovered to the baseline total, continue adjusting the next operating branch; when the total flow rate recovers to the baseline total, stop compensation adjustment, and record the valve position change of the currently adjusted branch, the valve position changes of the remaining operating branches, the sampled temperature values ​​of the heat exchange section outlet, and the time of adjustment completion to generate a forward adjustment record; after the forward adjustment is completed, the edge end restores the valve positions of each operating branch to the coded values ​​before adjustment, so that the system returns to the operating state before the disturbance; then, reduce the valve position of the currently adjusted branch by one control resolution unit, and according to the same flow rate percentage sorting results during the forward adjustment process, increase the control resolution unit of the remaining operating branches in sequence for flow compensation; when the sum of the flow rates of each operating branch recovers to the baseline total, stop reverse compensation, and record the reverse valve position change process, the branch flow rate change process, and the sampled temperature values ​​of the heat exchange section outlet to generate a reverse adjustment record;

[0078] Through the above-mentioned forward and reverse adjustment processes, two opposite disturbances, one increasing the flow rate and the other decreasing the flow rate, are generated in the same inlet branch. If only a single-direction adjustment is performed, when the temperature change is not obvious due to compensation in other operating branches, it is easy to cause the response characteristics of different heat exchange sections to be similar. Using two disturbances in opposite directions can distinguish the affected heat exchange sections by utilizing the difference in the direction of temperature change.

[0079] S2-2. At the edge end, the outlet temperature of each heat exchange segment at the start of the positive adjustment is extracted based on the positive adjustment record, and this temperature is used as the reference temperature for each heat exchange segment. Specifically, for the heat exchange segment numbered Hi, the outlet temperature Ti0 at the start of the positive adjustment is recorded, and the outlet temperature Ti(k) of this heat exchange segment is read according to a fixed sampling period, and the temperature difference at the current sampling point is calculated.

[0080] Ti(k)-Ti0;

[0081] Where k represents the sampling position number, and Ti(k) represents the outlet temperature collected at the k-th sampling position; by calculating the temperature difference instead of directly comparing the outlet temperatures, the initial temperature difference between different heat exchange sections will not affect the response judgment; at the edge end, the temperature difference sequence of each heat exchange section is traversed according to the sampling position order, and the position that first satisfies the same sign of the temperature difference between two adjacent valid sampling positions and the absolute value of the temperature difference of the later sampling position is greater than the absolute value of the temperature difference of the previous sampling position is determined as the starting position;

[0082] For example, if the temperature difference sequence of a heat exchange section is -0.1, -0.3, -0.5, then the second sampling bit satisfies the condition that the temperature difference direction is maintained and the change amplitude increases, and the second sampling bit is used as the starting point of the change; at the same time, the positive or negative sign of the temperature difference at this sampling bit is read as the change sign; among them, the sampling bit with a temperature difference of 0 does not participate in the sign judgment, so as to avoid determining the starting position in advance when the equipment has not generated an effective response;

[0083] When a sample bit is missing data, skip that sample bit and continue reading the next valid sample bit; after completing the positive adjustment response extraction, read the reverse adjustment record at the edge; starting from the start time of the reverse adjustment, compare the absolute values ​​of the temperature difference of each sample bit according to the sampling order; when the absolute value of the temperature difference is in the decreasing phase, continue reading the subsequent sample bits; when it is detected that the absolute value of the temperature difference changes from a decreasing state to an increasing state, the previous sample bit will be determined as the recovery bit;

[0084] For example, if the absolute temperature difference sequence after reverse adjustment is 0.6, 0.4, 0.2, 0.3, then the third sampling point is the position with the lowest temperature difference change, and the third sampling point is determined as the recovery point; if the absolute temperature difference drops to 0 and remains unchanged, then the sampling point that first reaches 0 is taken as the recovery point, in order to avoid the problem that the equipment recovery process cannot form a turning point;

[0085] S2-3. At the edge end, the forward adjustment records, reverse adjustment records, and temperature response characteristics corresponding to each water inlet branch are organized to generate a response sequence. Specifically, the heat exchange section number is used as the response sequence index, and the forward adjustment records, reverse adjustment records, change symbols, start change positions, and recovery positions corresponding to each heat exchange section during the adjustment process of the same water inlet branch are arranged in numerical order.

[0086] The response sequence includes at least the current inlet branch number, heat exchange section number, forward valve position change record, reverse valve position change record, change symbol, initiation change position number, and recovery position number; wherein, the forward adjustment record and reverse adjustment record are used to characterize the disturbance process generated by the current inlet branch on the heat exchange section, the change symbol is used to characterize the direction of temperature response, the initiation change position is used to characterize the response start position, and the recovery position is used to characterize the recovery position after the disturbance ends; after the edge end completes the generation of the response sequence, the response sequence is uploaded to the cloud; the cloud further performs matching calculations between the branch and the heat exchange section based on the uploaded multiple inlet branch response sequences;

[0087] Through the above implementation process, the edge end uses active valve position disturbance to replace manual judgment, so that the temperature response of different water inlet branches to the heat exchange section is transformed into response characteristics with direction, time location and recovery process, providing a data basis for subsequent identification of isolated objects;

[0088] In practical applications: For example, a low-temperature economizer contains multiple parallel heat exchange sections. When the maintenance worker points to the inlet branch J3, the edge end first reads the flow rate of all operating branches and calculates the baseline total. Then, the J3 valve position is increased by one control resolution unit, and the valve positions of other branches are successively decreased from high to low according to the proportion of flow rate before adjustment, so that the total flow rate is restored to the state before adjustment, forming a positive adjustment record. After restoring the valve position, the J3 valve position is decreased by one control resolution unit again, and the valve positions of other branches are increased in the same way to restore the total flow rate, forming a reverse adjustment record.

[0089] During the temperature response analysis, if the outlet temperature difference of heat exchange section H3 changes from 0 to -0.2 or -0.5 after positive adjustment, while the temperature difference changes of other heat exchange sections are small, then H3 is recorded as having a change position and a negative change sign. During the reverse adjustment, if the absolute value of the temperature difference of H3 decreases from 0.5 to 0.1 and then increases back to 0.2, then the sampling position before the decrease and increase is recorded as the recovery position. Finally, the response sequence corresponding to J3 is formed at the edge and uploaded to the cloud for subsequent determination of the response correlation between J3 and the target heat exchange section.

[0090] S3. The cloud uses the change symbol, start position and recovery position in the response sequence as matching features, performs secondary assignment under the constraint of the isolation diagram, and generates a branch and heat exchange section mapping matrix through projection iteration. The inlet branch with the target heat exchange section sorting number 1 and its connected return branch and flue gas damper are used as isolation objects, and the mapping value between the target heat exchange section and the inlet branch is output.

[0091] In this embodiment, in order to solve the problem that it is difficult to accurately determine the target isolation object due to the similar pipeline connection relationship between multiple parallel heat exchange sections of the low temperature economizer, after obtaining the disturbance response sequence of each water inlet branch, the cloud further calculates the response correlation between the water inlet branch and the heat exchange section by combining the actual medium path relationship in the isolation diagram.

[0092] This process does not directly determine the isolation target based on temperature changes. Instead, it simultaneously uses the response direction, response location, and recovery location to form a matching cost, and combines this with equipment topology constraints to filter out combinations without physical pathways. Then, it obtains the mapping relationship between branches and heat exchange sections through assignment calculations, and finally determines the inlet water branch, return water branch, and flue gas damper corresponding to the target heat exchange section. This implementation process includes the following steps:

[0093] S3-1. The cloud receives the response sequence uploaded by the edge terminal and reads the change symbol, start position and recovery position of the heat exchange section corresponding to each water inlet branch, and converts the response characteristics into the matching cost between the branch and the heat exchange section.

[0094] Specifically, for any candidate matching relationship consisting of any inlet branch and any heat exchange section, the cloud reads the change symbol, start position and recovery position generated during the adjustment process of the inlet branch, and compares them with the actual response characteristics of the heat exchange section.

[0095] The sign difference is used to represent the number of inconsistencies in the direction of temperature change. The cloud compares the change sign in the response sequence corresponding to the candidate inlet branch with the response sign of the current heat exchange section item by item. When the two signs are the same, it is recorded as 0, and when the two signs are different, it is recorded as 1. All comparison results are added together to obtain the sign difference.

[0096] For example, if a certain inlet branch response sequence contains 5 heat exchange section response symbols, 3 of which are different from the current heat exchange section symbol, then the symbol difference of this candidate combination is 3; the initiation difference is used to represent the difference in response start time position; the cloud reads the initiation position number in the response sequence corresponding to the candidate inlet branch and the initiation position number of the current heat exchange section response, subtracts the two numbers and takes the absolute value to obtain the initiation difference; the recovery difference is used to represent the difference in disturbance recovery position, the cloud reads the two recovery position numbers in the same way and calculates the absolute difference to obtain the recovery difference; after completing the calculation of the three feature differences, the cloud adds the symbol difference, the initiation difference and the recovery difference to obtain the matching cost between the current inlet branch and the heat exchange section:

[0097] Matching cost = Sign difference + Origin variation + Recovery difference;

[0098] The matching cost is used to represent the proximity between the response features of two objects. The smaller the value, the smaller the difference between the two in terms of perturbation direction, response start time and recovery process. S3-2, the cloud screens the candidate matching relationship according to the water inlet path relationship in the isolation diagram, and performs secondary assignment calculation based on the screened matching cost to generate the mapping matrix between the water inlet branch and the heat exchange section.

[0099] Specifically, the cloud reads the set of water inlet paths in the isolation diagram. For any combination of water inlet branch and heat exchange section, it determines whether the water inlet branch node can reach the corresponding heat exchange section node along the water inlet path. When there is no directed path from the water inlet branch to the heat exchange section, the matching cost data corresponding to the combination is deleted. When there is a water inlet path, the combination is retained for subsequent calculations. Through the above path filtering, erroneous matches that only have similar temperature responses but no actual medium connection relationship can be removed.

[0100] For example, if two heat exchange sections are in the same ambient temperature and their outlet temperature changes in similar directions, but one of the heat exchange sections is not connected to the current inlet branch, then this combination will not be included in the subsequent mapping calculation. After the candidate combination screening is completed, the cloud uses the assignment constraint that each inlet branch corresponds to only one heat exchange section and each heat exchange section corresponds to only one inlet branch, and forms a cost matrix from the matching costs of all candidate combinations, and performs secondary assignment calculation.

[0101] Specifically, the cloud first selects an initial matching relationship based on the cost matrix, establishing a one-to-one correspondence between all inlet branches and heat exchange sections. Then, it reads the sum of the matching costs for all combinations under the current matching relationship, using this as the current mapping cost. Based on the current mapping relationship, the cloud swaps the matching relationships between different inlet branches and heat exchange sections, and recalculates the mapping cost after the swap. If the swapped mapping cost is less than the current mapping cost, the swap result is retained; otherwise, the original matching relationship is maintained. Through multiple rounds of swapping and updating, the overall matching cost is reduced, generating the initial mapping matrix.

[0102] In the initial mapping matrix, rows represent inlet branch numbers, columns represent heat exchange section numbers, and matrix elements represent the matching relationship between the corresponding inlet branch and heat exchange section. After the initial mapping matrix is ​​generated, the cloud further adjusts the non-zero elements in the matrix according to the direction of decreasing matching cost.

[0103] Specifically, the matching cost corresponding to the current non-zero element is read. If adjusting an element can reduce the matching cost of the corresponding combination of the inlet branch and the heat exchange section, the adjusted element value is retained; otherwise, the element value before adjustment is restored. After the element adjustment is completed, the sum of the elements in each row and the sum of the elements in each column are calculated in the cloud. The matrix elements are then normalized by dividing them by the sum of the elements in their respective rows and columns. This ensures that the element relationships between different heat exchange sections corresponding to the same inlet branch and the element relationships between different inlet branches corresponding to the same heat exchange section are kept within the same calculation range, thus obtaining the final mapping matrix.

[0104] S3-3. The cloud determines the target inlet branch corresponding to the target heat exchange section based on the mapping matrix, and verifies it by combining response characteristics and path relationships;

[0105] Specifically, the cloud reads the matrix column corresponding to the target heat exchange section number, sorts all matrix elements in the column according to their values ​​from high to low, and reads the inlet branch corresponding to the matrix element with sorting number 1 as a candidate inlet branch; then, the cloud returns the isolation diagram to read the connection relationship corresponding to the candidate inlet branch, and obtains the return branch connected to the inlet branch and the flue gas damper that controls the flue gas flow of the path; to avoid incorrect selection due to relying solely on mapping values, the cloud further performs three verifications:

[0106] The first step is to determine whether there is an inlet path between the candidate inlet branch and the target heat exchange section. If there is no inlet path, it means that the branch cannot provide medium input to the target heat exchange section, and the verification result is invalid.

[0107] The second step is to determine whether the matrix element corresponding to the candidate inlet branch is still in the first position of the column sorting number of the target heat exchange section, in order to confirm that the current candidate branch still has the highest matching relationship.

[0108] The third step involves reading the change symbols in the response sequence of the candidate inlet branch and counting the number of symbols that are the same as those in the target heat exchange section. This number is then compared with the number of response points. When the two numbers are equal, it indicates that all valid response points are in the same direction. If all three judgments are true, the candidate inlet branch is determined as the target inlet branch. Otherwise, a mapping conflict result is generated, and the process is returned to the edge to re-execute the corresponding inlet branch disturbance acquisition process to update the response sequence.

[0109] S3-4. The cloud generates an isolation object based on the topological connection relationship of the target inlet branch and outputs a mapping value for subsequent state judgment. Specifically, the cloud reads the return branch and flue gas damper connected to the target inlet branch along the isolation diagram, and forms an isolation object by combining the target inlet branch, the corresponding return branch, and the corresponding flue gas damper. At the same time, the cloud reads the intersection value of the column where the target heat exchange section is located and the row where the target inlet branch is located in the mapping matrix, and uses this element value as the mapping value corresponding to the current isolation object. This mapping value is used to represent the degree of response correlation between the current inlet branch and the target heat exchange section, and serves as the initial correlation parameter in the subsequent isolation state probability calculation. After the isolation object is generated, the cloud reads the path relationship in the isolation diagram again to determine whether the inlet branch, return branch, and flue gas damper in the isolation object are all located in the same target heat exchange section isolation path.

[0110] Specifically, the cloud reads the connection nodes between the three isolation components and the target heat exchange section along the water inlet path, water return path, and flue gas path, respectively. When all three components belong to the same target heat exchange section corresponding to the path, it is determined that the isolation object meets the path constraint, and the isolation object and mapping value are sent to the edge end; otherwise, the current isolation object is discarded, and the next matching relationship is selected for calculation.

[0111] Through the above process, the cloud uses a combination of response feature calculation and equipment topology constraints to determine the real isolation object, avoiding deviations in the isolation object caused by similar temperature responses of adjacent heat exchange sections, adjacent equipment numbers, or manual selection errors. At the same time, it provides a mapping basis for subsequent isolation status judgment. In practical applications: for example, a certain low-temperature economizer includes heat exchange sections H1, H2, and H3, of which the target heat exchange section to be repaired is H3.

[0112] After the edge device perturbs multiple water inlet branches, it uploads the corresponding response sequences for each branch. The cloud computing system finds that the sign difference between water inlet branches J3 and H3 is 0, the starting position difference is 1, and the recovery position difference is 0, indicating that their matching cost is low. Although water inlet branch J2 has a similar temperature change direction, it is directly deleted because there is no water inlet path connecting to H3. Subsequently, the cloud computing system calculates that the matrix element ranking between J3 and H3 is the highest. After verifying the existence of the path, the ranking position, and the consistency of the sign of the change, J3 is identified as the target water inlet branch, and its connected return water branch R3 and flue gas damper D3 are read to form an isolation object. The cloud computing system further reads the matrix element value at the intersection of J3 and H3 as a mapping value and sends the isolation object and the mapping value to the edge device for subsequent recursive judgment of the isolation status.

[0113] S4. At the edge, the system performs flue gas isolation, inlet and outlet water cut-off, pressure relief, and liquid drainage on the isolated object, and collects the flow rate, pressure, and temperature difference of the target heat exchange section to generate a state sequence. At the cloud, the system performs Bayesian recursion based on the mapping value as a priori, and outputs the probability of success and failure of flow interruption, pressure loss, and temperature drop.

[0114] In this embodiment, in order to determine whether the maintenance requirements are met after the isolation action is performed, it is necessary to perform online analysis of the actual operating status after the isolation action. Since other heat exchange sections are still operating during the non-stop maintenance of the low-temperature economizer, it is not possible to determine whether the target heat exchange section has truly completed the media isolation based solely on the valve closure status.

[0115] Therefore, this implementation collects flow, pressure, and temperature change data of the target heat exchange section after performing isolation at the edge, and converts the sampled data into state change characteristics. Based on this, the cloud combines the mapping values ​​obtained in the previous steps as initial association information to perform progressive state probability updates for the flow interruption, pressure loss, and cooling processes, so that the isolation state judgment is simultaneously constrained by the target object matching relationship and actual operational changes. This implementation process includes the following steps:

[0116] S4-1. The edge terminal receives the isolation object sent from the cloud. The isolation object includes the target water inlet branch number, the target water return branch number, and the target flue gas damper number. The edge terminal reads the corresponding control interface according to the device number in the isolation object and outputs control commands in the order of flue gas isolation, water inlet cut-off, and water return cut-off.

[0117] Specifically, the edge terminal first sends a closing command to the target flue gas damper and reads the feedback status of the flue gas damper. After the flue gas damper is closed, it sends a closing command to the target inlet branch valve. After the inlet branch valve is closed, it sends a closing command to the target return branch valve. By executing the closing action according to the medium flow direction, the flue gas side and water side isolation actions are completed in the execution sequence of the process chain. During the isolation action, the edge terminal synchronously collects the operating status data of the target heat exchange section.

[0118] The inlet flow sensor collects the inlet medium flow rate of the target heat exchange section, the pressure sensor collects the internal pressure of the target heat exchange section, and the temperature sensors collect the inlet and outlet temperatures of the target heat exchange section. The edge end records the data values ​​corresponding to the sampling time according to a fixed sampling period and generates flow rate, pressure, and temperature difference sequences, respectively. The flow rate sequence represents:

[0119] F = {F1, F2, ..., Fk}

[0120] Pressure sequence representation:

[0121] P = {P1, P2, ..., Pk}

[0122] Temperature difference sequence representation:

[0123] T = {T1, T2, ..., Tk}

[0124] Where k represents the current sampling position number, Fk represents the inlet flow rate collected at the kth sampling position, Pk represents the internal pressure collected at the kth sampling position, and Tk represents the inlet and outlet temperature difference calculated at the kth sampling position; by generating the above state sequence, subsequent judgments are based on continuous state changes after the isolation action, rather than the detection results at a single moment.

[0125] S4-2. The edge terminal performs differential calculation on the state sequence, converting the change direction between consecutive sampling bits into state change bits, and statistically analyzing the degree of state change during flow interruption, pressure loss, and cooling based on the change direction; specifically, for the k-th sampling bit, the edge terminal reads the k-th sampling value and the (k-1)-th sampling value respectively, and performs differential calculation:

[0126] Changes in flow rate:

[0127] ΔFk=Fk-Fk-1;

[0128] Pressure change:

[0129] ΔPk = Pk - Pk-1;

[0130] Temperature difference change:

[0131] ΔTk = Tk - Tk-1; After the difference calculation is completed, the edge end generates the corresponding change position according to the sign of the change; when ΔFk is less than 0, a flow rate decrease position is generated; when ΔFk is greater than 0, a flow rate increase position is generated; when ΔFk is equal to 0, no direction change position is generated; pressure change position and temperature difference change position are generated using the same rules;

[0132] For example, during the isolation process, if the inlet flow rate of the target heat exchange section decreases from 100 m³ / h to 20 m³ / h, a flow rate decrease position is generated at the corresponding sampling point; if the flow rate subsequently increases again due to valve leakage, a flow rate increase position is generated at the corresponding sampling point; after the change position is generated, the number of flow rate, pressure, and temperature difference change positions are counted at the edge end; for the flow status, the flow interruption factor is obtained by subtracting the number of flow rate increase positions from the number of flow rate decrease positions.

[0133] For pressure conditions, the pressure drop factor is obtained by subtracting the pressure rise from the pressure drop.

[0134] For the temperature difference state, the cooling factor is obtained by subtracting the number of temperature difference rises from the number of temperature difference falls. The above factor is used to represent the trend of the current state changing towards the isolation target. When the number of temperature falls increases, the factor increases, indicating that the isolation process continues to advance. When the number of temperature rises increases, the factor decreases, indicating that there is a recovery trend in the isolation state.

[0135] S4-3. The cloud establishes an initial state probability relationship based on the mapping value obtained in S3, and performs Bayesian recursion in combination with the change factor uploaded by the edge to calculate the probability of the establishment and failure of the flow interruption, pressure loss and cooling states respectively. Specifically, the cloud reads the mapping value between the target heat exchange section and the target water inlet branch, and uses the mapping value as the initial probability information of the establishment of the isolation state; at the same time, it calculates the difference between 1 and the mapping value, and uses the difference as the initial probability information of the failure of the isolation state.

[0136] For example, the closer the mapping value is to 1, the higher the response correlation between the target inlet branch and the target heat exchange section, and the higher the initial probability of the flow interruption, depressurization and cooling states. The lower the mapping value, the weaker the correlation between the isolated object and the target heat exchange section, and the higher the initial probability of failure.

[0137] After the initial probability settings are completed, the cloud reads the interruption factor, pressure loss factor, and cooling factor uploaded by the edge terminal. For the interruption state, the cloud uses the number of traffic drop bits as evidence of establishment and the number of traffic rebound bits as evidence of destruction, and updates the interruption establishment probability and interruption destruction probability according to the probability relationship between the current sampling bit and the previous sampling bit.

[0138] Specifically, when a new flow rate decrease position is added to the current sampling position, the probability of flow interruption is increased; when a new flow rate increase position is added to the current sampling position, the probability of flow interruption failure is increased. After the flow interruption status is updated, the cloud reads the flow interruption status update result as the basis for pressure loss status calculation. Pressure loss probability update is only performed when the flow interruption status has a trend of establishment. For pressure loss status, the cloud updates the probability of pressure loss establishment and the probability of pressure loss failure based on the number of pressure decrease positions and the number of pressure increase positions. After the pressure loss status meets the establishment trend, the cloud further performs cooling status calculation and updates the probability of cooling establishment and the probability of cooling failure based on the number of temperature difference decrease positions and the number of temperature difference increase positions.

[0139] Through the above progressive calculation, the states of flow interruption, pressure loss and cooling are correlated step by step according to the actual isolation process, avoiding the direct judgment of isolation completion based on temperature changes when the target heat exchange section has not yet completed the flow interruption.

[0140] S4-4. The cloud performs state verification on the probability of establishment and the probability of failure for each state according to the execution order of flow interruption, pressure loss, and cooling, and generates an isolated state sequence or isolated change bit. Specifically, the cloud first reads the probability of flow interruption establishment and the probability of flow interruption failure. When the probability of flow interruption establishment is higher than the probability of flow interruption failure, it continues to read the probability of pressure loss. Otherwise, the current sampling bit is recorded as the abnormal sampling bit of flow interruption. After the flow interruption state meets the conditions, the cloud continues to determine whether the probability of pressure loss establishment is higher than the probability of pressure loss failure. When the conditions are met, it continues to determine whether the probability of cooling establishment is higher than the probability of cooling failure. When the probability of flow interruption establishment is higher than the probability of flow interruption failure, the probability of pressure loss establishment is higher than the probability of pressure loss failure, and the probability of cooling establishment is higher than the probability of cooling failure, the cloud assembles the flow state, pressure state, temperature difference state, and the probability of each state corresponding to the current sampling bit into an isolated state sequence and uploads it to the cloud database. When any state does not meet the above relationship, the cloud searches for the sampling bit whose establishment probability is not higher than the probability of failure for the first time according to the sampling order, and determines the sampling bit as the isolated change bit, and uploads the state data corresponding to the isolated change bit.

[0141] Through the above process, it is possible to determine whether the isolated object has truly formed a state of flow interruption, pressure loss and temperature drop under the environment of non-stop operation, and to locate the stage of isolation anomaly by the location of state change, so as to provide a basis for subsequent maintenance release or locking control;

[0142] Through the above implementation method, the real-time status is collected after the isolation action is completed at the edge. The cloud uses the mapping value to constrain the initial relationship of the state probability, and then combines the trends of flow, pressure and temperature difference to progressively update the isolation status. This makes the isolation judgment consider both the object matching relationship and the actual operation changes, avoiding the problem of wrong release caused by judging the isolation completion solely based on the valve action result.

[0143] In practical applications: For example, the isolation objects corresponding to the target heat exchange section H3 include flue gas damper D3, inlet water branch J3, and return water branch R3; at the edge end, D3 is closed first, followed by J3 and R3, and the inlet flow rate, internal pressure, and inlet-outlet temperature difference of H3 are continuously collected; during the sampling process, the inlet flow rate gradually decreases from the initial value, the pressure decreases synchronously, and the temperature difference gradually decreases. The cloud generates the initial state probability based on the mapping value between J3 and H3; if the number of flow rate decrease points increases during continuous sampling, but no obvious flow rate rebound point appears, the probability of flow interruption gradually increases; after flow interruption is established, the pressure decrease point continues to increase, and the probability of pressure loss continues to increase; subsequently, the temperature difference decrease point increases, and the probability of cooling increases; when the probability of the three states being established is higher than the corresponding failure probability, the cloud generates an isolation state sequence and allows subsequent maintenance release judgment; if a flow rebound occurs midway, causing the probability of flow interruption failure to exceed the establishment probability, the cloud records the first abnormal sampling point and stops subsequent state release judgment.

[0144] S5. Based on the mapping value and the probability of flow interruption, pressure loss and temperature drop, the cloud issues maintenance release instructions according to the process chain. When the flow interruption, pressure loss or temperature drop status fails, the subsequent maintenance process is locked from the node associated with the failure status along the process chain.

[0145] In this embodiment, in order to avoid premature entry into maintenance state during the online isolation process due to incorrect selection of isolation object, abnormal local state, or failure to meet the requirements of process execution sequence, it is necessary to perform step-by-step release control of maintenance process based on the mapping relationship obtained from the previous steps and the results of changes in flow interruption, pressure loss, and temperature drop.

[0146] In this implementation, the cloud does not directly issue maintenance commands based on the status of a single device. Instead, it simultaneously reads the relationships between isolated objects, the probability of isolation status, and the dependencies between processes, combining the status judgment results with the process chain to ensure that each maintenance action is executed only after the prerequisite isolation conditions are met. When an anomaly occurs in any link, the subsequent actions are locked by tracing back along the process chain based on the location of the anomaly, preventing the continued execution of maintenance operations that do not meet the conditions when the isolation status changes. This implementation process includes the following steps:

[0147] S5-1. The cloud reads the mapping value generated by S3 and the probability data of flow interruption, pressure loss and cooling status generated by S4, and converts the above data into maintenance release judgment parameters; specifically, the cloud reads the intersection element value of the row where the target water inlet branch is located in the matrix column corresponding to the target heat exchange section, and uses the element value as the mapping value corresponding to the current isolation object;

[0148] The mapping value represents the degree of response correlation between the current isolated object and the target heat exchange section, and it originates from the branch-to-heat exchange section mapping matrix. Simultaneously, the cloud reads the flow interruption, pressure loss, and cooling status data corresponding to the current sampling point. Each status includes a probability of establishment and a probability of failure. The flow interruption status includes the probability of flow interruption establishment and the probability of flow interruption failure; the pressure loss status includes the probability of pressure loss establishment and the probability of pressure loss failure; and the cooling status includes the probability of cooling establishment and the probability of cooling failure.

[0149] The cloud uses the mapping value as the object association parameter, the probability of each state being true as the state judgment parameter, and the probability of each state being broken as the anomaly judgment parameter, and forms object value, state value, and breakage value respectively. Among them, the object value is used to represent the matching relationship between the current isolated object and the target heat exchange section; the state value is used to represent the degree of change of the current isolation state towards maintenance conditions; and the breakage value is used to represent the degree of change of the current isolation state towards non-isolation.

[0150] For example, when the mapping value corresponding to a certain isolation object is high, and the probability of flow interruption, pressure loss, and temperature drop are all higher than the corresponding failure probability, it indicates that the current isolation object has a strong matching relationship with the target heat exchange section, and the isolation process is in a continuous completion state; otherwise, it is necessary to restrict subsequent maintenance actions.

[0151] S5-2. The cloud establishes the dependency relationship between maintenance actions based on the process chain generated in S1, and generates a process dependency sequence. Specifically, the cloud reads the node relationship in the process chain. Since the isolation process of the low-temperature economizer requires blocking the medium input first, then releasing the internal pressure, and finally reducing the equipment temperature, there is a sequential dependency relationship between each process. The cloud first reads the connection relationship between the flue gas isolation node, the water inlet cut-off node, the water return cut-off node, the pressure relief node, and the liquid discharge node, and uses the release status of the flue gas isolation node as a prerequisite judgment condition for the water inlet cut-off node and the water return cut-off node.

[0152] When the flue gas isolation node is in the release state of 1, it indicates that the flue gas isolation action is completed, and the water inlet cut-off node and the water return cut-off node are ready to execute. When the flue gas isolation node is in the release state of 0, the water inlet cut-off node and the water return cut-off node remain locked. Subsequently, the cloud uses the release states of the water inlet cut-off node and the water return cut-off node together as the prerequisite judgment condition for the pressure relief node. The pressure relief node enters the judgment state only after both the water inlet side and the water return side have completed the cut-off.

[0153] Furthermore, the cloud uses the release status of the pressure relief node as a prerequisite for the liquid discharge node. Once the internal pressure is released, the liquid discharge node enters the execution state. Through the above dependency relationship, a process dependency sequence consisting of flue gas isolation, water inlet and outlet cut-off, pressure relief, and liquid discharge is generated. If multiple process nodes are at the same execution level, parallel judgment is performed based on the incoming edge relationship in the process chain.

[0154] For example, if both the inlet water cut-off and the return water cut-off are triggered by the flue gas isolation node, then both nodes must meet the flue gas isolation release conditions before they can be executed.

[0155] S5-3. The cloud performs maintenance release judgment based on object value, status value and process dependency sequence, and issues maintenance release instructions according to process sequence;

[0156] Specifically, the cloud first determines whether the currently isolated object is the inlet branch with sorting number 1 corresponding to the target heat exchange section. During the determination, it reads the column of the mapping matrix where the target heat exchange section is located, sorts the mapping values ​​of the corresponding inlet branch, and if the inlet branch corresponding to the current isolated object is located at sorting number 1, the object determination result is true; otherwise, the object determination result is false. Subsequently, the cloud reads the probability of flow interruption and the probability of flow interruption failure, the probability of pressure loss and the probability of pressure loss failure, and the probability of cooling failure and the probability of cooling failure, and performs status determination for each.

[0157] Specifically:

[0158] When the probability of a flow interruption being established is higher than the probability of a flow interruption being destroyed, the result of the flow interruption status is recorded as 1.

[0159] When the probability of pressure loss is higher than the probability of pressure loss failure, the pressure loss state result is recorded as 1;

[0160] When the probability of cooling is successful is higher than the probability of cooling failure, the cooling state result is recorded as 1.

[0161] Otherwise, the corresponding status result is recorded as 0; after the status result is generated, the cloud executes and calculates according to the process dependency sequence;

[0162] For example, the release criteria for the flue gas isolation process include: the object judgment result is 1;

[0163] The flow interruption status meets the requirements; when both conditions are met simultaneously, the flue gas isolation node release status is set to 1, and a flue gas isolation maintenance release command is sent to the edge end; after the flue gas isolation node release status is 1, the inlet water cut-off node and return water cut-off node are judged; only when the current process status result is met and all the preceding process release statuses are 1, the current process release status is set to 1; otherwise, the current process release status remains 0; by judging process by process, the maintenance action is released in the order of flue gas isolation, inlet and return water cut-off, pressure relief and liquid drainage, so as to avoid the premature execution of subsequent actions when the preceding isolation conditions are not met;

[0164] S5-4. The cloud performs lock propagation based on the abnormal process nodes and performs step-by-step release judgment after the state is restored. Specifically, when the release status of any process node is 0, the cloud reads the first node in the process chain with a release status of 0 and determines the node as an abnormal process node. Subsequently, the cloud traverses all associated nodes after the abnormal process node along the outgoing edge of the process chain and sets the release status of the traversed process nodes to 0, while generating a lock command.

[0165] For example, when a pressure relief node's release status becomes 0 due to an abnormal pressure loss, the cloud traverses the process chain backwards through the drainage nodes, locking them to prevent drainage operations from being performed when the pressure condition is not met. During subsequent sampling, the cloud continuously reads the status results corresponding to the abnormal process nodes. When the abnormal process node is detected to have recovered for the first time, it does not immediately unlock all nodes, but instead reads the preceding process nodes level by level along the incoming edges of the process chain. Specifically, the cloud sequentially determines the release status of each process node preceding the abnormal process node:

[0166] If the release status of all preceding process nodes is 1, then the current abnormal node and subsequent nodes are unlocked.

[0167] If the release status of any preceding process node is 0, the locked state is maintained, and the process continues to wait for the preceding conditions to be restored. Through the above-mentioned lock propagation and step-by-step release method, the maintenance process can be dynamically adjusted according to the changes in the isolation status, avoiding the direct restoration of all processes during the abnormal recovery process, which would cause erroneous execution.

[0168] Through the above implementation method, the cloud uses mapping values ​​to limit the selection range of isolation objects, and combines the probability of flow interruption, pressure loss and temperature drop to judge the actual degree of isolation. Then, it controls the release of maintenance actions according to the process dependency relationship, realizing closed-loop control from isolation confirmation to maintenance execution. When the isolation state changes, the locking relationship is propagated along the process chain to stop subsequent maintenance actions after the abnormal node, reducing the risk of misoperation in the online maintenance process.

[0169] In practical applications: For example, after the target heat exchange section H3 completes the closure of flue gas damper D3, inlet water branch J3, and return water branch R3, the cloud reads the mapping value between J3 and H3 and the probability of the current flow interruption, pressure loss, and cooling status; if the mapping value is ranked first in the corresponding inlet water branch of the target heat exchange section, and the probability of flow interruption is higher than the probability of flow interruption failure, the probability of pressure loss is higher than the probability of pressure loss failure, and the probability of cooling is higher than the probability of cooling failure, then the cloud will first allow the flue gas isolation process;

[0170] Subsequently, when the flue gas isolation node is in the release state of 1, the inlet water cut-off and return water cut-off nodes enter the judgment; after both nodes meet the state conditions, the pressure relief node obtains the execution conditions; if an abnormal pressure loss state is detected during the pressure relief process, causing the pressure relief node to change to the release state of 0, the cloud immediately locks the drainage node; when subsequent sampling finds that the pressure loss state has recovered, the cloud re-checks the status of the upstream nodes such as flue gas isolation and inlet / return water cut-off, and only releases the drainage lock when all upstream nodes are in the release state, so that the entire maintenance process is adjusted according to the actual changes in the isolation state.

[0171] Furthermore, the online isolation operation and maintenance management platform based on the modular cryogenic economizer includes:

[0172] The isolation modeling module is used to construct an isolation diagram in the cloud based on the target heat exchange section number in the maintenance work order, as well as the connection relationship of each heat exchange section, inlet branch, return branch and flue gas damper, and generate a process chain according to the sequence of flow interruption, pressure loss, liquid discharge and cooling.

[0173] The response generation module is used to apply the same amplitude reverse valve position adjustment to the inlet branch pointed to by the maintenance work order and the inlet branch with an inlet path to the target heat exchange section in the isolation diagram when the edge end keeps running in the other heat exchange sections and the total inlet flow remains unchanged. It generates a response sequence according to the change sign, start position and recovery position of the outlet temperature of each heat exchange section and uploads it to the cloud.

[0174] The object matching module is used in the cloud to perform secondary assignment under the constraint of the isolation graph by using the change symbol, start position and recovery position in the response sequence as matching features. It generates a branch and heat exchange section mapping matrix through projection iteration, and uses the inlet branch with the target heat exchange section sorting number 1 and its connected return branch and flue gas damper as isolation objects, and outputs the mapping value between the target heat exchange section and the inlet branch.

[0175] The state determination module is used at the edge to perform flue gas isolation, inlet and outlet water cut-off, pressure relief and liquid drainage on the isolated object, and collect the flow rate, pressure and temperature difference of the target heat exchange section to generate a state sequence; the cloud performs Bayesian change point recursion with the mapping value as a priori, and outputs the probability of success and failure of flow interruption, pressure loss and temperature drop;

[0176] The process control module is used to issue maintenance release instructions in the cloud according to the mapping value and the probability of flow interruption, pressure loss and temperature drop. When the flow interruption, pressure loss or temperature drop status fails, the subsequent maintenance process is locked along the process chain from the node associated with the failure status.

[0177] Working principle: This solution starts with the online shutdown of the target heat exchange section for maintenance. First, the cloud reads the target heat exchange section number and the connection relationship between each heat exchange section, inlet branch, return branch and flue gas damper of the low temperature economizer from the maintenance work order. It generates the isolation path of the target heat exchange section along the inlet, return and flue gas passages, and forms a process chain according to the dependency relationship between flue gas isolation, inlet and return water cut-off, pressure loss, liquid discharge and cooling. Then, at the edge end, under the condition that the other heat exchange sections are still running and the total inlet flow rate is restored to the level before adjustment, equal amplitude forward and reverse valve position adjustment is performed one by one for each candidate inlet branch with an inlet path to the target heat exchange section, and the change sign, start position and recovery position of the outlet temperature of each heat exchange section are recorded.

[0178] The cloud platform converts these response characteristics into matching costs, performs secondary assignment and projection iteration under isolation path constraints, determines the actual associated inlet branch of the target heat exchange section, and then obtains the associated return branch and flue gas damper along the isolation diagram to form the isolation object and mapping value. The edge end closes the flue gas damper, inlet branch valve and return branch valve according to the process chain, and performs pressure relief and liquid drainage, while collecting the inlet flow, internal pressure and inlet-outlet temperature difference of the target heat exchange section. The cloud platform uses the mapping value as the initial association basis, and recursively infers the probability of establishment and failure of flow, pressure loss and temperature drop based on the decrease and increase positions of flow, pressure and temperature difference, and combines the probability results with the preceding relationships in the process chain to issue maintenance release instructions level by level. When the failure probability of any state is not lower than the establishment probability, the cloud platform locks the maintenance processes that have not yet been executed from the process node bound to that state until the state is restored and all preceding processes are still in the release state.

[0179] In practical applications, the heat exchange section H3 of a certain low-temperature economizer needs to be repaired during unit operation. The cloud first obtains the inlet water branch, return water branch, and flue gas passage leading to H3 from the equipment connection data, and generates the execution sequence of flue gas damper closure, inlet and return water cut-off, pressure loss, liquid discharge, and cooling. The edge terminal adjusts the candidate inlet water branches J1, J2, and J3 in sequence, increasing the valve position control resolution unit by one unit each time, and then decreasing the same unit after restoration, while simultaneously adjusting other operating branches to restore the total inlet water flow. If H3 first shows a temperature change in the opposite direction after J3 is adjusted, and the starting position and restoration position match the connection path of H3, the cloud determines J3 as the target inlet water branch through assignment calculation, and obtains the return water branch R3 and flue gas damper D3 as isolation objects.

[0180] After closing D3, J3, and R3 at the edge, H3 is depressurized and drained. The cloud system sequentially confirms the flow interruption, pressure loss, and temperature drop states based on the sampled changes in H3 inlet flow, internal pressure, and inlet / outlet temperature difference. When all three states are met, interface inspection, interface loosening, and subsequent maintenance procedures are released according to the process chain. If the pressure rises again during the depressurization period, the probability of pressure loss damage exceeds the probability of pressure loss being met. The cloud system immediately locks the draining and subsequent maintenance procedures. The lock is released only after the pressure drops again and the pre-isolation state is still met.

[0181] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An online isolation operation and maintenance management method based on a modular cryogenic economizer, characterized in that, include: S1. The cloud platform constructs an isolation diagram based on the target heat exchange section number in the maintenance work order, as well as the connection relationship between each heat exchange section, inlet branch, return branch and flue gas damper, and generates a process chain according to the sequence of flow interruption, pressure loss, liquid drainage and cooling. S2. When the edge end continues to operate in the other heat exchange sections and the total inlet flow remains unchanged, apply the same amplitude reverse valve position adjustment to the inlet branch pointed to by the maintenance work order and the inlet branch with an inlet path to the target heat exchange section in the isolation diagram one by one. Generate a response sequence according to the change sign, start position and recovery position of the outlet temperature of each heat exchange section and upload it to the cloud. S3. The cloud uses the change symbol, start position and recovery position in the response sequence as matching features, performs secondary assignment under the constraint of the isolation diagram, and generates a branch and heat exchange section mapping matrix through projection iteration. The inlet branch with the target heat exchange section sorting number 1 and its connected return branch and flue gas damper are used as isolation objects, and the mapping value between the target heat exchange section and the inlet branch is output. S4. At the edge, the system performs flue gas isolation, inlet and outlet water cut-off, pressure relief, and liquid drainage on the isolated object, and collects the flow rate, pressure, and temperature difference of the target heat exchange section to generate a state sequence. At the cloud, the system performs Bayesian recursion based on the mapping value as a priori, and outputs the probability of success and failure of flow interruption, pressure loss, and temperature drop. S5. Based on the mapping value and the probability of flow interruption, pressure loss and temperature drop, the cloud issues maintenance release instructions according to the process chain. When the flow interruption, pressure loss or temperature drop status fails, the subsequent maintenance process is locked along the process chain from the node associated with the failure status.

2. The online isolation operation and maintenance management method based on a modular cryogenic economizer according to claim 1, characterized in that: S1 includes: S1-1, The cloud uses the heat exchange section number, water inlet branch number, water return branch number and flue gas damper number as point identifiers, and the water flow direction or flue gas flow direction between two points as the edge direction. It searches backward from the target heat exchange section to the water inlet header, water return header and flue gas inlet respectively, and generates the water inlet path, water return path and flue gas path composed of point identifiers and the connection order of adjacent points. S1-2. Form isolation groups by combining the inlet branches in the inlet path, the return branches in the return path, and the flue gas baffles in the flue gas path. Remove the points and connected edges in the isolation groups one by one, and determine whether there are still directed paths from the inlet header, return header, and flue gas inlet to the target heat exchange section. Keep the isolation groups where all three types of directed paths are cut off. When the number of points in the isolation groups is the same, sort them in ascending order according to the total number of edges from each point to the target heat exchange section, and take the isolation group with the first position in the sort to generate the isolation diagram. S1-3. The process is directed by the closure of the flue gas damper to the closure of the inlet branch and the return branch, and the closure of the inlet branch and the return branch together to the depressurization. The depressurization leads to the liquid discharge, and the liquid discharge leads to the cooling. A directed edge for the process is established. In each round, the process with zero incoming edges is selected to generate the current process position. The outgoing edges of the current process position are deleted and the selection continues until a process chain arranged in the order of execution is generated.

3. The online isolation operation and maintenance management method based on a modular cryogenic economizer according to claim 2, characterized in that: S2 includes: S2-1. At the edge end, the sum of the flow rates of each operating branch before adjustment is used as the baseline total. First, the valve position of the branch pointed to by the work order is increased by one control resolution unit. Then, the valve positions are decreased sequentially according to the flow rate ratio of the remaining operating branches before adjustment, until the sum of the flow rates of each operating branch is restored to the baseline total, generating a positive adjustment record. After restoring each operating branch to the valve position before adjustment, the valve position of the branch pointed to by the work order is decreased by one control resolution unit. Then, the valve positions are increased sequentially according to the flow rate ratio of the remaining operating branches before adjustment, until the sum of the flow rates of each operating branch is restored to the baseline total, generating a reverse adjustment record. S2-2. Using the outlet temperature of each heat exchange section at the start of the forward adjustment as the reference temperature, calculate the difference between the outlet temperature and the reference temperature according to a uniform sampling interval. The first sampling position that satisfies the same sign for two adjacent temperature differences and the absolute value of the latter is greater than the absolute value of the former is taken as the starting position, and the sign of the temperature difference at the starting position is taken as the change sign. From the start of the reverse adjustment, compare the absolute values ​​of the temperature difference according to the sampling order, and take the sampling position before the absolute value of the temperature difference changes from decreasing to increasing as the recovery position. S2-3. Arrange the forward adjustment records, reverse adjustment records, change symbols, start position and recovery position according to the heat exchange section number, generate the response sequence and upload it to the cloud.

4. The online isolation operation and maintenance management method based on a modular cryogenic economizer according to claim 3, characterized in that: S3 includes: S3-1. The cloud takes the change symbol, start change position and recovery position in the response sequence as input, calculates the number of inconsistent change symbols, the difference of start change position sequence number and the difference of recovery position sequence number between the water inlet branch and the heat exchange section as sign difference, start change difference and recovery difference respectively, and uses the sum of sign difference, start change difference and recovery difference as matching cost. S3-2. Based on the water inlet path in the isolation diagram, remove the combination of water inlet branch and heat exchange section that does not have a water inlet path. With each water inlet branch associated with one heat exchange section and each heat exchange section associated with one water inlet branch as a constraint, perform secondary assignment calculation on the matching cost of the remaining combinations to generate an initial mapping matrix. Update the non-zero elements in the initial mapping matrix according to the decreasing direction of the matching cost, and normalize the updated matrix elements by dividing them by the sum of the elements in their row and the sum of the elements in their column, respectively, to generate a mapping matrix.

5. The online isolation operation and maintenance management method based on a modular cryogenic economizer according to claim 4, characterized in that: S3 further includes: S3-3. Read the column containing the target heat exchange section in the mapping matrix, sort the mapping values ​​in the column in descending order, determine the inlet branch corresponding to the mapping value with the sorting number 1 as the candidate inlet branch, and read the return water branch and flue gas damper connected to the candidate inlet branch along the isolation diagram; determine whether there is an inlet path between the candidate inlet branch and the target heat exchange section, whether the sorting number of the mapping value is 1, and whether the number of consistent change signs in the response sequence is equal to the number of response points. When all three judgment results are true, the candidate inlet branch is determined as the target inlet branch; otherwise, a mapping conflict result is generated and the response sequence is regenerated at the edge end. S3-4. The return water branch and flue gas damper connected to the target water inlet branch are combined with the target water inlet branch to form an isolation object, and the matrix element value at the intersection of the column where the target heat exchange section is located and the row where the target water inlet branch is located is determined as the mapping value; according to the isolation diagram, it is verified whether the water inlet branch, return water branch and flue gas damper in the isolation object are in the same target heat exchange section isolation path. When the three are in the same isolation path, the isolation object and the mapping value are sent to the edge end.

6. The online isolation operation and maintenance management method based on a modular cryogenic economizer according to claim 5, characterized in that: S4 includes: S4-1. The edge end outputs a shutdown command to the flue gas damper, the inlet branch valve, and the return branch valve in sequence according to the flue gas damper number, the inlet branch valve number, and the return branch valve number in the isolated object. It also collects the inlet flow rate, internal pressure, and inlet-outlet temperature difference of the target heat exchange section and generates the flow rate sequence, pressure sequence, and temperature difference sequence according to the sampling time. S4-2. Subtract the (k-1)th sample value from the kth sample value in the flow rate sequence, pressure sequence, and temperature difference sequence to obtain the flow rate change, pressure change, and temperature difference change, respectively. Generate the flow rate change bit, pressure change bit, and temperature difference change bit according to the signs of the flow rate change, pressure change, and temperature difference change. Subtract the number of rising bits from the number of falling bits as the flow interruption factor, pressure loss factor, and temperature drop factor, respectively.

7. The online isolation operation and maintenance management method based on a modular cryogenic economizer according to claim 6, characterized in that: S4 further includes: S4-3. Using the mapping value as the prior to the establishment and the mapping value minus one as the prior to the destruction, the number of falling positions in the flow interruption factor, pressure loss factor, and cooling factor are used as the evidence of establishment and the number of rising positions are used as the evidence of destruction. Bayesian recursion is performed on flow interruption, pressure loss, and cooling in sequence to generate the probability of flow interruption and the probability of flow interruption destruction, the probability of pressure loss and the probability of pressure loss destruction, and the probability of cooling and the probability of cooling destruction. S4-4. Read the success probability and failure probability of each according to the order of flow interruption, pressure loss and cooling. When the success probability of flow interruption is higher than the failure probability of flow interruption, the success probability of pressure loss is higher than the failure probability of pressure loss and the success probability of cooling is higher than the failure probability of cooling, upload the isolation state sequence to the cloud. Otherwise, determine the first sampling bit whose success probability is not higher than the failure probability as the isolation change bit and upload it to the cloud.

8. The online isolation operation and maintenance management method based on a modular cryogenic economizer according to claim 7, characterized in that: S5 includes: S5-1: Read the mapping value from the cloud and read the probability of the current interruption state, the pressure loss state, and the cooling state at the current sampling point; combine the mapping value, the probability of success, and the probability of failure to form the object value, the state value, and the failure value, respectively. S5-2. Based on the directed edge relationships between flue gas isolation, water inlet / outlet cut-off, pressure relief, and liquid discharge in the process chain, the release state of the flue gas isolation node is used as the prerequisite state of the water inlet / outlet cut-off node, the release state of the water inlet / outlet cut-off node is used as the prerequisite state of the pressure relief node, and the release state of the pressure relief node is used as the prerequisite state of the liquid discharge node, thus generating a process dependency sequence.

9. The online isolation operation and maintenance management method based on a modular cryogenic economizer according to claim 8, characterized in that: The S5 also includes: S5-3. Using the following as the inputs for release calculation: the probability of the mapping value belonging to the inlet branch with the corresponding sorting number 1 of the target heat exchange section, the probability of the flow interruption being higher than the probability of the flow interruption failure, the probability of the pressure loss being higher than the probability of the pressure loss failure, and the probability of the cooling being higher than the probability of the cooling failure; the results of each state being established are executed and calculated according to the process dependency sequence; when the state result corresponding to the current process is established and the release state of the previous process is 1, the release state of the current process is set to 1 and a maintenance release command is issued; otherwise, the release state of the current process is set to 0. S5-4. Read the first process node with a release status of 0, and traverse each node along the outgoing edge of the process chain. Set the release status of the traversed process node to 0 and generate a locking command. When the corresponding status of the first abnormal process node in the subsequent sampling bits is restored to true, check the release status of the preceding process node along the incoming edge of the process chain. Only unlock when all preceding processes have a release status of 1.

10. An online isolation operation and maintenance management platform based on a modular cryogenic economizer, characterized in that: include: The isolation modeling module is used to construct an isolation diagram in the cloud based on the target heat exchange section number in the maintenance work order, as well as the connection relationship of each heat exchange section, inlet branch, return branch and flue gas damper, and generate a process chain according to the sequence of flow interruption, pressure loss, liquid discharge and cooling. The response generation module is used to apply the same amplitude reverse valve position adjustment to the inlet branch pointed to by the maintenance work order and the inlet branch with an inlet path to the target heat exchange section in the isolation diagram when the edge end keeps running in the other heat exchange sections and the total inlet flow remains unchanged. It generates a response sequence according to the change sign, start position and recovery position of the outlet temperature of each heat exchange section and uploads it to the cloud. The object matching module is used in the cloud to perform secondary assignment under the constraint of the isolation graph by using the change symbol, start position and recovery position in the response sequence as matching features. It generates a branch and heat exchange section mapping matrix through projection iteration, and uses the inlet branch with the target heat exchange section sorting number 1 and its connected return branch and flue gas damper as isolation objects, and outputs the mapping value between the target heat exchange section and the inlet branch. The state determination module is used at the edge to perform flue gas isolation, inlet and outlet water cut-off, pressure relief and liquid drainage on the isolated object, and collect the flow rate, pressure and temperature difference of the target heat exchange section to generate a state sequence; the cloud performs Bayesian change point recursion with the mapping value as a priori, and outputs the probability of success and failure of flow interruption, pressure loss and temperature drop; The process control module is used to issue maintenance release instructions in the cloud according to the mapping value and the probability of flow interruption, pressure loss and temperature drop. When the flow interruption, pressure loss or temperature drop status fails, the subsequent maintenance process is locked along the process chain from the node associated with the failure status.

Images