Intelligent service system of smart community

Abstract

The application provides a kind of intelligent service system of wisdom community, it is related to wisdom community information field, including: service registration module, for the standardization encapsulation of multi-format heterogeneous service;Request analysis module, for the structural analysis and service unit matching of resident service request and governance decision event message respectively, obtain standardization arrangement request;Arrangement execution module, for service node to carry out availability verification, according to the type of dependence Relationship Mark selection combination arrangement mode drives each service unit coordination execution;Scheduling optimization module, for aggregating and calculating service node level operation index according to statistical period, determine the comprehensive health degree score of each service node through health degree evaluation model, while the weight coefficient of health degree evaluation model is adaptively updated.The application realizes from heterogeneous service standardization encapsulation to scheduling strategy continuous adaptive update, can maintain stable service supply capacity under the condition of dynamic change of format composition.

Description

Technical Field

[0001] This invention relates to the field of smart community information technology, and in particular to an intelligent service system for smart communities. Background Technology

[0002] With the continuous advancement of urbanization, the daily needs of community residents are becoming increasingly diverse. This has led to the emergence of various service models coexisting in communities, including charging, parking, elderly care, childcare, catering, housekeeping, and express delivery. Simultaneously, the field of community governance has created an urgent need for digital tools, with resident deliberations, community voting, and surveys requiring information technology support. However, these services are operated by different providers with varying service interface standards, data formats, business processes, and responsiveness. Unified access and collaborative management of these diverse services has become a prominent challenge in the construction of smart community information systems.

[0003] Existing smart community systems have significant shortcomings in multi-service integration. On the one hand, existing systems typically only handle one type of service, failing to unify the management of facility-related services with lifestyle services and governance participation services. On the other hand, existing systems generally employ static, pre-set scheduling rules, lacking the ability to automatically adjust scheduling strategies based on the actual operational status of each service. When a service node experiences a decrease in supply capacity or malfunction, the system can only rely on manual intervention, impacting residents' service experience and the platform's continued availability. Furthermore, existing systems typically handle resident deliberations, community voting, and other governance participation behaviors at the information collection level, making it difficult to automatically link governance decisions with specific service execution tasks, resulting in a disconnect between deliberation and implementation.

[0004] Chinese invention patent application CN118657646A discloses a distributed smart community platform and its communication method based on a zero-trust network. This platform provides identity authentication, access control, and load balancing capabilities at the network infrastructure layer through a zero-trust mechanism. Based on this, it builds business modules such as community health checks, community monitoring, service operation, and public participation, integrating functions such as community security equipment monitoring, property management work order management, and resident opinion collection. However, this platform cannot meet the unified orchestration and scheduling needs of multi-service resources in a community setting, nor can it adjust scheduling based on node operational status. Summary of the Invention

[0005] In view of this, the present invention provides an intelligent service system for smart communities to solve the technical problems of existing smart community systems lacking support for heterogeneous service providers in multiple business formats and being unable to dynamically generate and continuously optimize scheduling strategies based on the actual operating status of service nodes. It realizes the transformation from standardized encapsulation of heterogeneous services to continuous adaptive updates of scheduling strategies, enabling the intelligent service system to maintain a stable service supply capacity under the condition of dynamic changes in business format composition.

[0006] The technical solution of this invention is implemented as follows: On one hand, the present invention provides an intelligent service system for smart communities, comprising: The service registration module is used to standardize and encapsulate multi-business heterogeneous services, including facility services and lifestyle services, using protocol conversion adapters and field mapping rules. The standard response latency benchmark value, group identifier, and initial allocation weight of each encapsulated service are written into the service registration directory to form a service resource pool. The request parsing module is used to perform structured parsing and service unit matching on resident service requests and governance decision event messages based on the service resource pool, so as to obtain standardized orchestration requests. The orchestration execution module is used to receive standardized orchestration requests, read the current scheduling policy, perform availability verification on the service nodes involved in the standardized orchestration requests, select the combined orchestration mode according to the dependency relationship type to drive the coordinated execution of each service node, record the call timestamp and execution status of each service node, count the actual response latency of each service node, and synchronize the operation data containing the actual response latency and execution status statistics of each service node to the scheduling optimization module. The scheduling optimization module is used to aggregate and calculate service node-level operational indicators according to statistical periods, determine the comprehensive health score of each service node through a health assessment model, generate a structured scheduling strategy based on the comprehensive health score and write it into the scheduling strategy storage, and adaptively update the weight coefficients of the health assessment model based on the exponentially weighted moving average tracking volume and the dual-threshold bidirectional adjustment rule.

[0007] Based on the above technical solutions, preferably, the steps of the service registration module are as follows: For facility-type services, the communication protocol type of IoT devices is identified, the raw data streams reported by the devices are extracted, the raw data streams are encapsulated according to the standard event message structure, and the service quality level is assigned to them according to the urgency of the event data; the encapsulation result includes a unique stream identifier, timestamp sequence, device physical identifier, service event type field, real-time status field, and trigger command format specification; For lifestyle services, the original interface documents of the service providers are parsed. Based on the original interface documents, the interface protocol type, request parameter field list, response field list, and authentication method are extracted. A two-way mapping rule is established between the original fields and the standard fields of the intelligent service system, and the interface authentication method is configured. The two-way mapping rule converts the parameter fields of the standard calling protocol of the intelligent service system into personalized request parameters that conform to the service provider's interface definition according to the mapping rule in the request direction. In the response direction, the response fields returned by the service provider are converted into the standard response format of the intelligent service system. The conversion content covers parameter name mapping, data type conversion, and field merging processing. All business services submit their registrations using a unified community service unit structure and are written into the service registration directory.

[0008] Based on the above technical solutions, preferably, the request parsing module includes a governance state machine and a parser, and the specific steps of the request parsing module include: The governance state machine generates a globally unique deliberation identifier for each deliberation proposal. The process proceeds sequentially through five stages: solicitation of support, consultation and discussion, voting, public announcement of results, and follow-up on implementation, thereby obtaining information on governance decision-making events. After receiving the governance decision event message from the governance state machine, the parser encodes it according to the event type. To find the corresponding execution task template in the governance task template library for the retrieval key, extract field values ​​from the structured digest of the governance task template library according to the field mapping rules in the task template, and obtain the service unique identifier of the corresponding service unit in the service registry directory. and the API call address; among which the governance task template library is The primary key pre-stores a list of service type identifiers required for each type of meeting, the dependency types between service units, and a structured summary. The mapping rules from each field in the code to the service call parameters; The outputs of the governance state machine and the parser are encapsulated into a standardized orchestration request, including a request identifier. The system includes a trigger mode flag, an ordered set of service unit identifiers, a dependency type flag, and a request priority level. The request priority is carried in the request message or generated by mapping the urgency of governance decisions.

[0009] Based on the above technical solution, preferably, the governance state machine needs to transition between each state. If the transition condition is met, it automatically advances to the next state; if the condition is not met, it is set to the corresponding termination state. The transition conditions are as follows: The conditions for transitioning from supporting the solicitation to consultation and discussion are as follows: ,in The number of supporters. Minimum number of support personnel required for operations management; The conditions for transitioning from negotiation and discussion to voting are: ,in To determine the duration of the negotiations, The maximum negotiation time limit configured for the operations management party; The process of proceeding from the voting deadline to the results announcement must simultaneously meet the following conditions: the voting deadline has passed and... ,in, Indicates the approval rate of the votes in favor. This indicates the total number of valid affirmative votes. This represents the total number of valid votes. This indicates the threshold for the approval rate of proposals configured by the operations management team; When the voting deadline arrives The proposal state machine is set to the rejected and terminated state, and no governance decision event message is sent to the parser; Residents' real-name objections will be accepted starting from the result announcement period, and the cumulative number of valid objections during the announcement period will be counted. ,like If the state machine is set to the public withdrawal termination state and no governance decision event message is sent to the parser, then... The threshold for the proportion of objections triggered during the public notice period to be withdrawn is configured for the operation and management party; The conditions for transitioning from the result announcement state to the execution tracking state are: and Send a message carrying to the parser , Voting Results and Structured Summary News about governance decision-making events, among which The public notice period has been ongoing. To ensure the shortest possible public notice period; When the parser is If the search key does not find a corresponding template in the governance task template library, then Set the template configuration to a paused state and push it to the operations management team. Enumerated values, descriptive text of agenda item types, and A configuration request with already filled fields; After the operations management team completes the corresponding template configuration, Restore to the results public state, re-execute template retrieval and orchestration request encapsulation, add new template records to the governance task template library version management and increment the version number.

[0010] Based on the above technical solutions, preferably, the specific steps of the orchestration execution module include: Read the scheduling policy version number from the scheduling policy store. The corresponding scheduling strategy, wherein the scheduling strategy records the service-unique identifier of each service node. The scheduling status and weight allocation are as follows: the scheduling status includes normal status, degraded status and offline status; the weight allocation is dynamically updated by the scheduling optimization module according to the health score, and the initial value is determined by the service registration module during registration according to the average score of the same group. Verify the scheduling status of each node in the ordered set of service unit identifiers in the standardized orchestration request: When a service node is in an offline state, and there is a service group identifier that is the same as that node... Among the other available nodes, the one with the highest current assigned weight is selected as the replacement node; When a service node is in a degraded scheduling state, its assigned weight is reduced. When a service node is in a normal scheduling state, it is executed in a queue according to the priority level of the request in the standardized orchestration request, where the priority is carried by the resident request message or generated by the urgency of the governance decision. The actual set of nodes involved in this execution and the version number of the scheduling strategy. Write the request identifier The corresponding execution logs are used to select the orchestration method based on dependency type to drive the execution of each service node, and a call start timestamp is recorded for each service node. With return timestamp The actual response latency of the computing node : ; When each service node completes execution, the actual response latency and execution status are written to the execution log in real time. The execution status includes execution success, execution failure, timeout, compensation success, compensation business pending processing, compensation retry exhausted, or compensation timeout.

[0011] Based on the above technical solutions, preferably, the arrangement method includes sequential arrangement, parallel arrangement, and conditional branch arrangement, wherein, The sequential orchestration method sends call requests to the interface call addresses of each service node in sequence, using the return result of the previous node as the input parameter for the next node. If a node exceeds a preset timeout period... If no response is received, its execution status is set to timeout failure. The parallel orchestration sends a call request to all service nodes marked as parallel orchestration in this request, and collects the return results of each node in an asynchronous waiting manner. After all nodes return successfully, the return results are summarized. If any node fails to execute, the execution fails and a compensation rollback is triggered. The conditional branch orchestration execution pre-service unit compares its return value with the preset branch condition set one by one, and enters the branch path that meets the condition to execute the corresponding sequential orchestration or parallel orchestration sub-process. The branch path that is not selected does not trigger any service unit call.

[0012] Based on the above technical solutions, preferably, the compensation rollback specifically includes: Read the list of service nodes that have been successfully executed from the execution log, based on the compensation type flag of each node in the service registry. and compensation time window Calculate the remaining compensation period for each node in the list. : ; in The system moment when compensation is triggered. For the first Each service unit records the timestamp of the positive execution completion return in the execution log. This represents an enumeration value of a completely reversible compensation type. An enumeration of compensation types that are irreversible but can be compensated by business operations; sort the list Compensation operations are triggered sequentially after sorting in ascending order: The service unit is directly marked as having exceeded the compensation period and a special alarm is pushed to the operation and management party; The node calls the compensation operation interface to undo the positive execution result. If the call fails, it retryes by increasing the retry interval exponentially in a backoff manner until the maximum number of retries is reached. If the process still fails, the execution status is marked as exhausted due to compensation retries. The node calls the compensation operation interface, records an exception flag, and initiates the business compensation process. Upon successful call, the node's execution status is updated to "compensation business pending" and written to the execution log. If the call fails, it is retried using an exponential backoff method. Second-rate.

[0013] Based on the above technical solutions, preferably, the specific steps of the scheduling optimization module include: Statistical period As a time window, in each statistical period At the end, perform aggregate calculations on the operational data, based on the statistical period number. Identify each period, It is a positive integer; For the service registry entry Each service node, based on the actual response latency of all successful calls within this statistical period. The arithmetic mean of the actual average response delay is calculated. If no successful call occurs within this cycle, the current cycle will continue. ; Based on the number of successful calls within this period and number of failed calls Calculate the service availability rate ; The average resident evaluation score is calculated based on the arithmetic mean of the valid resident evaluation scores within this period. ; Will , , As input to the health assessment model, a comprehensive health score is calculated. ; Based on the comprehensive health score and the preset lower limit threshold of normal status and the lower limit threshold of the weighted state The relationships between these relationships determine the scheduling status of each service node, where : when The scheduling status is normal and the assigned weights remain unchanged; when When the scheduling state is reduced in weight, the reduction magnitude is based on the relative ranking factor within the group. Sure: ; in, For nodes Among all nodes in the same group identifier that are in normal or degraded status, sort by The ranking number when arranged from smallest to largest. This represents the total number of currently available nodes within the same group identifier; when When the scheduling status is offline, the assigned weight is cleared to zero, and the remaining available nodes under the same group identifier perform weight normalization. The updated scheduling status and assigned weights are written into the scheduling policy storage, and the weight coefficients of the health assessment model are adaptively updated.

[0014] Based on the above technical solutions, the preferred formula for calculating the comprehensive health score is: ; ; in For the first The service node is at the The overall health score for each statistical period. , , The response latency, service availability, and resident ratings were respectively measured in the [number]th month. Weighting coefficients for each statistical period The full score for resident evaluation; A piecewise mapping function for the response delay ratio, For response delay, high-quality gain coefficient, This represents the standard response latency baseline value preset during the registration of the i-th node.

[0015] More preferably, the adaptive updating of the weight coefficients of the health assessment model specifically includes: In each statistical period At the end, the original abnormal signal observations for the response latency dimension, service availability dimension, and resident evaluation were calculated respectively. ; For each dimension Maintain the index-weighted moving average tracking volume separately : ; in, It is the exponential smoothing coefficient. This represents the exponentially weighted moving average tracking value of the j-th dimension calculated and stored at the end of the (k-1)-th statistical period; according to With the enhanced trigger upper threshold and recovery trigger threshold Compare and calculate the weight adjustment increment for each dimension. : when hour, That is, for the first The weight coefficients for each dimension are increased by a fixed step size. ; when and hour, That is, for the first The weight coefficients of each dimension are reduced by a fixed step size. ; when The weights remain unchanged.

[0016] The present invention has the following advantages over the prior art: (1) By integrating heterogeneous services of multiple business formats into a unified manner and combining them with operational data feedback optimization through the service registration module, request parsing module, orchestration execution module and scheduling optimization module, the system realizes the transformation from standardized encapsulation of heterogeneous services to continuous adaptive updates of scheduling strategies, enabling the intelligent service system to maintain a stable service supply capacity under the condition of dynamic changes in business format composition. (2) The service registration module has designed protocol conversion and bidirectional field mapping rules for facility services and life services respectively, and encapsulates and registers them through a unified community service unit structure. This enables heterogeneous services with different communication protocols and different interface specifications to have a consistent description form and calling method within the system. New business service providers do not need to modify the core scheduling logic when they access the system, which reduces the expansion cost of the intelligent service system. (3) The orchestration execution module supports three orchestration modes: sequential, parallel and conditional branch. When execution fails, a classification compensation mechanism based on the remaining compensation time is introduced. The node with the shortest remaining compensation time window is processed first, which reduces the probability of time delay due to improper compensation operation order and reduces the number of abnormal scenarios requiring manual intervention. (4) The scheduling optimization module calculates the health score of the service node based on three types of operational indicators: response latency, service availability and resident evaluation through the health assessment model. It determines the weight reduction range by combining the relative ranking factor within the group, so that nodes with relatively better performance in the same group retain a reasonable share in the scheduling. At the same time, it adaptively updates the weight coefficients of each dimension through the index-weighted moving average tracking volume and the double threshold bidirectional adjustment rule, so that the health assessment model can automatically improve the sensitivity of the corresponding dimension when the operational problems are prominent, and automatically restore the equilibrium state after the problems subside. This effectively suppresses the degradation problem of the assessment model caused by the long-term unidirectional shift of the weight coefficients. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a method for an intelligent service system for a smart community according to the present invention; Figure 2 This is a flowchart of the service registration module of an intelligent service system for a smart community according to the present invention; Figure 3 This is a flowchart illustrating the state transition process of the governance state machine of an intelligent service system for a smart community, as described in this invention. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] like Figure 1 As shown, the present invention provides an intelligent service system for smart communities, comprising: The service registration module is used to standardize and encapsulate multi-business heterogeneous services, including facility services and lifestyle services, using protocol conversion adapters and field mapping rules. The standard response latency benchmark value, group identifier, and initial allocation weight of each encapsulated service are written into the service registration directory to form a service resource pool. The request parsing module is used to perform structured parsing and service unit matching on resident service requests and governance decision event messages based on the service resource pool, so as to obtain standardized orchestration requests. The orchestration execution module is used to receive standardized orchestration requests, read the current scheduling policy, perform availability verification on the service nodes involved in the standardized orchestration requests, select the combined orchestration mode according to the dependency relationship type to drive the coordinated execution of each service node, record the call timestamp and execution status of each service node, count the actual response latency of each service node, and synchronize the operation data containing the actual response latency and execution status statistics of each service node to the scheduling optimization module. The scheduling optimization module is used to aggregate and calculate service node-level operational indicators according to statistical periods, determine the comprehensive health score of each service node through a health assessment model, generate a structured scheduling strategy based on the comprehensive health score and write it into the scheduling strategy storage, and adaptively update the weight coefficients of the health assessment model based on the exponentially weighted moving average tracking volume and the dual-threshold bidirectional adjustment rule.

[0021] This invention integrates heterogeneous services from multiple business formats into a unified manner and combines them with operational data feedback optimization through a service registration module, a request parsing module, an orchestration and execution module, and a scheduling optimization module. This enables the intelligent service system to maintain a stable service supply capacity under the condition of dynamic changes in business formats by achieving standardized encapsulation of heterogeneous services and continuous adaptive updates of scheduling strategies.

[0022] like Figure 2 As shown, in one embodiment of the present invention, the steps of the service registration module are as follows: For facility-type services, identify the communication protocol type of IoT devices, extract the raw data streams reported by the devices, encapsulate the raw data streams according to the standard event message structure, and assign service quality levels to them based on the urgency of the event data; the encapsulation result includes a unique stream identifier, timestamp sequence, device physical identifier, service event type field, real-time status field, and trigger command format specification, and the communication protocol type includes MQTT, CoAP, or HTTP; For lifestyle services, the original interface documents of the service providers are parsed. Based on the original interface documents, the interface protocol type, request parameter field list, response field list, and authentication method are extracted. A two-way mapping rule is established between the original fields and the standard fields of the intelligent service system, and the interface authentication method is configured. The two-way mapping rule converts the parameter fields of the standard calling protocol of the intelligent service system into personalized request parameters that conform to the service provider's interface definition according to the mapping rule in the request direction. In the response direction, the response fields returned by the service provider are converted into the standard response format of the intelligent service system. The conversion content covers parameter name mapping, data type conversion, and field merging processing. All business services submit their registrations using a unified community service unit structure and are written into the service registration directory.

[0023] Understandably, facility-related services correspond to IoT device access scenarios, including charging piles, access control, parking lot gates, public cameras, etc.; lifestyle-related services correspond to commercial services provided by business service providers with REST or SOAP interfaces, including catering, housekeeping, express delivery, childcare, elderly care, etc.

[0024] Each community service unit must include the following full metadata during registration: Service Unique Identifier Service capability description text, trigger method flag (event-driven or request-driven), standard response latency baseline value. (The unit is milliseconds, provided by the service provider during registration, and used as a reference benchmark for the response latency item in the health assessment of the scheduling optimization module.) This includes the node's global ID in the service registry (unique and unchanging during system operation), current availability status flag, service interface call address, and identifier of the service group it belongs to. and compensation type flag and the address of the compensation operation interface. The service type is assigned by the operations management party during the registration review process. Each node in the group provides the same type of service and can be substituted for another in scheduling; the initial weights assigned to nodes in the same group. The number of nodes within the group is initialized evenly, i.e. ,in For the same The total number of service nodes in the group. This weight is dynamically updated by the scheduling optimization module based on the health score during subsequent operation.

[0025] Compensation type flag Take enumeration value or one. This indicates that the positive execution result of the service can be completely reversed through business operations, corresponding to lifestyle services such as order cancellation, payment refund, and permission revocation; This indicates that the positive execution result of the service is irreversible at the physical level, but the subsequent impact can be eliminated through business compensation operations. For facility-related IoT services, such as charging has been completed but the billing record can be marked as abnormal and a manual refund can be initiated, or access control has been granted but the access record can be marked as abnormal and a security alert can be triggered. The fields are declared by the business service provider during registration, and the intelligent service system confirms them manually by the operation and management party during the registration review. In case of any discrepancy between the two, the value confirmed by the operation and management party shall prevail. The service unit must also be written into the compensation time window simultaneously. , which represents the maximum time interval that the compensation operation interface can accept after the i-th service node completes its forward execution, in seconds. It is declared by the business service provider and confirmed by the operation management party. If the service provider does not provide it, the operation management party will configure the default value according to the industry practice of this business type.

[0026] In one embodiment of the present invention, the service event type field distinguishes three categories: telemetry data, event alarms and control commands. The urgency level of emergency event data is marked as the highest priority, and the end-to-end latency requirement is less than 50 milliseconds. The urgency level of routine monitoring data is marked as medium priority. The urgency level of historical backtracking data is marked as low priority.

[0027] This invention's service registration module designs protocol conversion and bidirectional field mapping rules for facility-type services and lifestyle-type services respectively, and encapsulates and registers them through a unified community service unit structure. This enables heterogeneous services with different communication protocols and interface specifications to have a consistent description and invocation method within the system. All registered service units in the service registration directory together constitute the system's service resource pool. When a new business service provider joins, it only needs to complete the encapsulation and submit the registration through the corresponding type of adapter, without modifying the core scheduling logic. The service registration directory is automatically updated, thereby supporting the continuous expansion of community service types and reducing the expansion cost of the intelligent service system.

[0028] In one embodiment of the present invention, the request parsing module includes a governance state machine and a parser, and the specific steps of the request parsing module include: The governance state machine generates a globally unique deliberation identifier for each deliberation proposal. The process proceeds sequentially through five stages: solicitation of support, consultation and discussion, voting, public announcement of results, and follow-up on implementation, resulting in a governance decision event message. The proposal includes the type of matter (selected from the system's pre-set list of types, such as public facility maintenance, community environmental improvement, and public service procurement), a description of the proposal, the location involved, the estimated cost range, and the identity of the proposer. After receiving the governance decision event message from the governance state machine, the parser encodes it according to the event type. To find the corresponding execution task template in the governance task template library for the retrieval key, extract field values ​​from the structured digest of the governance task template library according to the field mapping rules in the task template, and obtain the service unique identifier of the corresponding service unit in the service registry directory. and the API call address; among which the governance task template library is The primary key pre-stores a list of service type identifiers required for each type of meeting, the dependency types between service units, and a structured summary. The mapping rules from each field in the code to the service call parameters; The outputs of the governance state machine and the parser are encapsulated into a standardized orchestration request, including a request identifier. The system includes a trigger mode flag, an ordered set of service unit identifiers, a dependency type flag, and a request priority level. The request priority is carried in the request message or generated by mapping the urgency of governance decisions.

[0029] Understandably, when residents submit forms, they need to review the structured summary. Each field undergoes a pre-validation of its validity, with the validation rules corresponding to the item type code. Bind storage in the governance task template library.

[0030] Specifically, for the location field, its value is verified to be non-empty and consistent with the valid address format in the community geographic information database; for the estimated cost range field, it is verified to be a valid numerical range expression and its upper limit does not exceed the single proposal cost approval limit configured in the intelligent service system. (Unit: Yuan, configured and written into the system parameter table by the operations management party). If any required field fails validation, the intelligent service system returns a field error message to the resident, refuses to generate a deliberation identifier, and requires the resident to correct and resubmit. After the field pre-validation passes, the intelligent service system generates a globally unique deliberation identifier for this deliberation proposal. and for that Instantiate a state progression instance independently.

[0031] The request parsing module of this invention combines a governance state machine with a parser. It establishes a mapping relationship between the type encoding of deliberation items and service call parameters through a governance task template library. This enables the governance decision results to automatically trigger the orchestration and execution of the corresponding services after the conditions for transition between states are met, thus eliminating the information gap between governance deliberation and execution.

[0032] like Figure 3 As shown, in one embodiment of the present invention, the governance state machine needs to transition between states. If the transition condition is met, it automatically advances to the next state; if the condition is not met, it is set to the corresponding termination state. The transition conditions are as follows: The conditions for transitioning from supporting the solicitation to consultation and discussion are as follows: ,in The number of supporters. Minimum number of support personnel required for operations management; The conditions for transitioning from negotiation and discussion to voting are: ,in To determine the duration of the negotiations, The maximum negotiation time limit configured for the operations management party; The process of proceeding from the voting deadline to the results announcement must simultaneously meet the following conditions: the voting deadline has passed and... ,in, Indicates the approval rate of the votes in favor. This indicates the total number of valid affirmative votes. This represents the total number of valid votes. This indicates the threshold for the approval rate of proposals configured by the operations management team; When the voting deadline arrives The proposal state machine is set to the rejected and terminated state, and no governance decision event message is sent to the parser; Residents' real-name objections will be accepted starting from the result announcement period, and the cumulative number of valid objections during the announcement period will be counted. ,like If the state machine is set to the public withdrawal termination state and no governance decision event message is sent to the parser, then... The threshold for the proportion of objections triggered during the public notice period to be withdrawn is configured for the operation and management party; The conditions for transitioning from the result announcement state to the execution tracking state are: and Send a message carrying to the parser , Voting Results and Structured Summary News about governance decision-making events, among which The public notice period has been ongoing. To ensure the shortest possible public notice period; When the parser is If the search key does not find a corresponding template in the governance task template library, then Set the template configuration to a paused state and push it to the operations management team. Enumerated values, descriptive text of agenda item types, and A configuration request with already filled fields; After the operations management team completes the corresponding template configuration, Restore to the results public state, re-execute template retrieval and orchestration request encapsulation, add new template records to the governance task template library version management and increment the version number.

[0033] Understandably, when the duration of negotiation meets the requirements... If valid participation records have been received during the consultation and discussion period (i.e., at least one resident has submitted their consultation opinion), the state machine automatically advances to the voting state; if the consultation lasts for a certain period of time... If the number of residents actively participating in the discussion area is zero, meaning no residents submitted opinions during the discussion phase, the system will terminate the discussion. The corresponding deliberation state is set to the "negotiation invalidation termination" state. No governance decision event message is sent to the parser, but a notification of negotiation invalidation and a suggestion to modify the proposal content are pushed to the proposal initiator. The "negotiation invalidation termination" state, along with the "withdrawal termination," "rejection termination," and "public withdrawal termination" states, belongs to the final state that does not trigger orchestration execution. Once any of the above-mentioned termination states are reached, no further state transition operations will be accepted. The historical process record will be retained by the system for residents to review and for the operation and management to audit.

[0034] Through the aforementioned dual-mode parsing mechanism, both request-driven resident service needs and event-driven governance decision execution tasks are normalized to the same data structure before entering the orchestration and execution module. This allows the subsequent orchestration and execution module to process them uniformly without distinguishing the trigger source, reducing the coupling complexity of the core execution logic of the intelligent service system. At the same time, the full-process quantitative jump condition design of the governance state machine also embeds the compliance of the deliberation process into the technical logic of the intelligent service system.

[0035] Furthermore, the governance state machine also handles the following abnormal paths: In the solicitation and consultation / discussion phases, the proposer can submit a withdrawal request. The intelligent service system verifies the proposer's identity. Once the identity of the proposer matches, the governance state machine instance is set to the withdrawn and terminated state, and a withdrawal notice is sent to all residents who have supported the proposal. After entering the voting state, the proposer's unilateral right to withdraw automatically becomes invalid. After the governance state machine transitions to the execution trace state, the parser uses... To retrieve the key, perform an exact key-value lookup in the governance task template library. If the lookup returns null, the intelligent service system will then... Set the template configuration to a paused state and push the template containing this information to the operations management backend. Enumerated values, item type description text, and structured summary The configuration request for fields with filled content has been submitted; once the operations management team completes the corresponding template configuration, the intelligent service system will... The system restores the results to the public disclosure state, re-executes the template retrieval and arrangement request encapsulation, adds new template records to the governance task template library version management with auto-incrementing version numbers, and records the configuration time and the configuration person's identifier for subsequent auditing. This mechanism ensures that the deliberation and voting results completed by residents are not discarded due to missing templates.

[0036] In one embodiment of the present invention, the specific steps of the orchestration execution module include: Read the scheduling policy version number from the scheduling policy store. The corresponding scheduling strategy, among which, The latest policy version number written by the scheduling optimization module in the policy storage is a positive integer, incremented each time the scheduling optimization module updates the scheduling policy. The scheduling policy records the service-unique identifier of each service node. The scheduling status and weight allocation are as follows: the scheduling status includes normal status, degraded status and offline status; the weight allocation is dynamically updated by the scheduling optimization module according to the health score, and the initial value is determined by the service registration module during registration according to the average score of the same group. Verify the scheduling status of each node in the ordered set of service unit identifiers in the standardized orchestration request: When a service node is in an offline state, and there is a service group identifier that is the same as that node... Among the other available nodes, the one with the highest current assigned weight is selected as the replacement node; When a service node is in a degraded scheduling state, its assigned weight is reduced. When a service node is in a normal scheduling state, it is executed in a queue according to the priority level of the request in the standardized orchestration request, where the priority is carried by the resident request message or generated by the urgency of the governance decision. The actual set of nodes involved in this execution and the version number of the scheduling strategy. Write the request identifier The corresponding execution logs are used to select the orchestration method based on dependency type to drive the execution of each service node, and a call start timestamp is recorded for each service node. With return timestamp The actual response latency of the computing node : ; When each service node completes execution, the actual response latency and execution status are written to the execution log in real time. The execution status includes execution success, execution failure, timeout, compensation success, compensation business pending processing, compensation retry exhausted, or compensation timeout.

[0037] Understandable, start timestamp For the first i The system time at which this node issued the request. The system time at which the response is received for this call to the i-th node.

[0038] When a service node is in a downgraded scheduling state, it still participates in the actual execution of the current request and is not replaced by the system. However, the orchestration execution module marks the current scheduling state of the node as downgraded in the operational data reported to the scheduling optimization module after the current request is completed. At the end of the next statistical period, the scheduling optimization module calculates and updates the weight allocation based on the latest health score of the node using the downgrade formula. The new weight allocation after downgrading takes effect from the next scheduling policy version. That is, the effect of the downgrade operation is reflected in the traffic allocation ratio of the next scheduling period, rather than interrupting the currently being processed request.

[0039] In one embodiment of the present invention, the arrangement methods include sequential arrangement, parallel arrangement, and conditional branch arrangement, wherein: The sequential orchestration method sends call requests to the interface call addresses of each service node in sequence, using the return result of the previous node as the input parameter for the next node. If a node exceeds a preset timeout period... If no response is received, its execution status is set to timeout failure. The parallel orchestration sends a call request to all service nodes marked as parallel orchestration in this request, and collects the return results of each node in an asynchronous waiting manner. After all nodes return successfully, the return results are summarized. If any node fails to execute, the execution fails and a compensation rollback is triggered. The conditional branch orchestration execution pre-service unit compares its return value with the preset branch condition set one by one, and enters the branch path that meets the condition to execute the corresponding sequential orchestration or parallel orchestration sub-process. The branch path that is not selected does not trigger any service unit call.

[0040] Understandable, preset timeout duration The settings can be configured according to the actual usage scenarios of the community service system, and this invention does not impose specific limitations on them.

[0041] This invention supports three orchestration modes: sequential, parallel, and conditional branching. When execution fails, it introduces a classification compensation mechanism based on the sorting of remaining compensation time. The node with the shortest remaining compensation time window is processed first, which reduces the probability of time delay due to improper order of compensation operations and reduces the number of abnormal scenarios requiring manual intervention.

[0042] In one embodiment of the present invention, compensation rollback specifically includes: Read the list of service nodes that have been successfully executed from the execution log, based on the compensation type flag of each node in the service registry. and compensation time window Calculate the remaining compensation period for each node in the list. : ; in The system moment when compensation is triggered. For the first Each service unit records the timestamp of the positive execution completion return in the execution log. This represents an enumeration value of a completely reversible compensation type. An enumeration of compensation types that are irreversible but can be compensated by business operations; sort the list Compensation operations are triggered sequentially after sorting in ascending order: The service unit is directly marked as having exceeded the compensation period and a special alarm is pushed to the operation and management party; The node calls the compensation operation interface to undo the positive execution result. If the call fails, it retryes by increasing the retry interval exponentially in a backoff manner until the maximum number of retries is reached. If the process still fails, the execution status is marked as exhausted due to compensation retries. The node calls the compensation operation interface, records an exception flag, and initiates the business compensation process. Upon successful call, the node's execution status is updated to "compensation business pending" and written to the execution log. If the call fails, it is retried using an exponential backoff method. Second-rate.

[0043] Understandable, for The node, because of its All are arranged in the same order. After the node, multiple The nodes are arranged in reverse order of their original execution.

[0044] The timeliness priority mechanism minimizes the remaining compensation time window. Nodes are prioritized for processing, effectively reducing the probability that subsequent nodes will exceed the time limit due to the time occupied by preceding compensation operations; the classification compensation processing mechanism clearly distinguishes between physically irreversible and reversible operations at the technical level, reducing the total amount of manual intervention, while retaining complete records of abnormal situations for post-event auditing.

[0045] In one embodiment of the present invention, the specific steps of the scheduling optimization module include: Statistical period As a time window, in each statistical period At the end, perform aggregate calculations on the operational data, based on the statistical period number. Identify each period, It is a positive integer; For the service registry entry Each service node, based on the actual response latency of all successful calls within this statistical period. The arithmetic mean of the actual average response delay is calculated. If no successful call occurs within this cycle, the current cycle will continue. : ; in, This represents the total number of times the i-th node was successfully invoked within the k-th statistical period. Only calls that received a response are counted, excluding calls that timed out or failed. This represents the summation index of all successful calls to the i-th service node within the k-th statistical period. This represents the actual response latency of the i-th node during the g-th call in this statistical period; Based on the number of successful calls within this period and number of failed calls Calculate the service availability rate : ; in, This indicates the total number of failed calls (including timeouts) within the current period, derived from the execution status statistics field in the execution log; The average resident evaluation score is calculated based on the arithmetic mean of the valid resident evaluation scores within this period. : ; in, This indicates the total number of valid resident evaluations within the current period. This represents the summation index of all valid resident evaluations for the i-th service node within the k-th statistical period. This represents the score of the valid resident evaluation in item e, with a value range of [value missing]. , This represents the maximum score for resident evaluation; Will , , As input to the health assessment model, a comprehensive health score is calculated. ; Based on the comprehensive health score and the preset lower limit threshold of normal status and the lower limit threshold of the weighted state The relationships between these relationships determine the scheduling status of each service node, where : when The scheduling status is normal and the assigned weights remain unchanged; when When the scheduling state is reduced in weight, the reduction magnitude is based on the relative ranking factor within the group. Sure: ; ; in, For nodes Among all nodes in the same group identifier that are in normal or degraded status, sort by The ranking number when arranged from smallest to largest. This represents the total number of nodes in the same group that are either in a normal or downgraded state. This represents the adjusted weight of node i after calculation using the weight reduction formula. Indicates the initial weight allocation. This represents the maximum weight reduction ratio, with a value range of (0,1), and is configured by the operations management team. when When the scheduling status is offline, the assigned weight is cleared to zero, and the remaining available nodes under the same group identifier perform weight normalization. The updated scheduling status and assigned weights are written into the scheduling policy storage, and the weight coefficients of the health assessment model are adaptively updated.

[0046] Understandably, nodes that rank lower within the group generally perform worse. The smaller, The larger the value, the greater the reduction in weight; the reduction in weight for relatively well-performing nodes within the group is appropriately narrowed, preserving the best available alternative nodes within the group to handle more scheduling traffic. Specifically, when... hour, , , The weighted formula degenerates into This means that the weight reduction adjustment will not be performed when there is only one node in the group.

[0047] After being demoted, for the same The set of all currently available nodes in the group except node i Execution weight normalization: ; in The normalized adjusted weight for the g-th available node in the same group; The current assigned weight of this node before the normalization operation; denominator The sum of the current assigned weights of all available nodes in the same group (excluding node i); multiplied by Make the same after the weight reduction The sum of the weights of all nodes in the group (including node i itself) is exactly equal to 1, ensuring that the weighted scheduling allocation ratio of the orchestration execution module remains effective regardless of any node state change. When this happens, the node's scheduling status is set to offline. Clear to zero, and then normalize the remaining available nodes in the same group according to the above normalization formula. Substitute the weights into the normalized weights.

[0048] In one embodiment of the present invention, the formula for calculating the comprehensive health score is as follows: ; ; in For the first The service node is at the The overall health score for each statistical period. , , The response latency, service availability, and resident ratings were respectively measured in the [number]th month. Weighting coefficients for each statistical period The full score for resident evaluation; A piecewise mapping function for the response delay ratio, For response delay, high-quality gain coefficient, This represents the standard response latency baseline value preset during the registration of the i-th node.

[0049] Understandable. Furthermore, each weighting coefficient is strictly greater than its lower protection limit. When When the actual delay is not better than the benchmark, the function value equals It itself is linearly increasing in the interval (0,1); when The time delay indicates that the actual delay is better than the benchmark. The function value is in logarithmic form with β as the gain coefficient and truncated to the upper limit of 1. The range of β is... Configured by the operations management team; functions in The continuity is satisfied ( ),exist The segment grows smoothly with diminishing marginal returns, and is ensured by a hard cap. The range of is strictly constrained within (0,1], thus making ... The value range is strictly (0,1).

[0050] In one embodiment of the present invention, the weight coefficients of the health assessment model are adaptively updated, specifically including: In each statistical period At the end, the original abnormal signal observations for the response latency dimension, service availability dimension, and resident evaluation were calculated respectively. : ; ; ; ; ; in, Indicates the first The original abnormal signal observations in the response delay dimension within a statistical period, Indicates the first The total number of service nodes in an available state (normal or downgraded) within a statistical period; Indicates the first The set of service nodes that are available within a statistical period, with the number of elements in the set being [number]. ; To find the sum index, traverse... All available nodes in the set; Indicates the first The service node is at the The actual average response time within a statistical period Indicates the first The cross-node arithmetic mean of the actual average response latency of all available nodes globally within a statistical period. Indicates the first The original abnormal signal observations in the service availability dimension within each statistical period, with values ​​being non-negative integers. Indicates the first The sum of the number of compensation operations triggered by all available service nodes globally within a statistical period. Indicates the first The service node is at the The number of times the compensation operation was triggered within a statistical period. Indicates the first The original abnormal signal observations of the resident evaluation dimension within a statistical period. Indicates the first The cross-node arithmetic mean of the average resident evaluation scores for all available service nodes globally within a statistical period. Indicates the first The service node is at the Average score of residents' evaluation within a statistical period; Indicates the first The arithmetic mean of the global resident evaluation average score across nodes within each statistical period is calculated and stored by the system at the end of the previous statistical period. Calculation of each statistical period Read directly; when the system is in the first statistical period ( )hour, Take initial value ; For each dimension Maintain the index-weighted moving average tracking volume separately : ; in, It is the exponential smoothing coefficient. This represents the exponentially weighted moving average tracking value of the j-th dimension calculated and stored at the end of the (k-1)-th statistical period; according to With the enhanced trigger upper threshold and recovery trigger threshold Compare and calculate the weight adjustment increment for each dimension. : when hour, That is, for the first The weight coefficients for each dimension are increased by a fixed step size. ; when and hour, That is, for the first The weight coefficients of each dimension are reduced by a fixed step size. ; when The weights remain unchanged.

[0051] Understandable, among which The step size for weight adjustment has a range of values. This value is configured by the operations management team in the system parameter table, with a default value of 0.05. This is the lower limit for the weighting coefficient protection, and its value range is [value range missing]. Configured by the operations management team, its default value is 0.1. Its purpose is to prevent the weight of any dimension from approaching zero due to continuous reduction. For the... After the incremental calculations for each dimension are completed, the update is executed. Then, the weight coefficients of the three dimensions were uniformly normalized.

[0052] Strengthen the upper threshold of triggering middle This means that the threshold is dynamically updated at the end of each statistical period, and the update rule is: in the first... At the end of each statistical period, the original abnormal signal observation values ​​of that period are used. Perform exponentially weighted smoothing on the threshold of the previous period, i.e. ,in This is the threshold smoothing coefficient, ranging from (0,1), configured by the operations management team, with a default value of 0.8. A larger value results in a slower and more stable response to short-term abnormal signals, while a smaller value makes the threshold more sensitive. During the system's initial runtime ( ), Initial thresholds for each dimension are determined and pre-configured by the operations management team in the system parameter table based on service scale and historical experience. If the operations management team does not provide these thresholds, the system will automatically use the pre-configured thresholds. As the default initial value, where The data collected by the system during the warm-up period is the first... The sample mean of the original abnormal signal in each dimension. Recover the trigger threshold. It remains a fixed constant throughout the entire system operation, configured by the operations management team in the system parameter table, and the system requires... This ensures that the two thresholds always maintain an effective high-low spacing relationship, preventing oscillations in the adjustment logic.

[0053] This value reflects the overall dispersion of the response capabilities of each service node within the current statistical period. When the response latency of all nodes tends to be consistent, this value approaches zero; when the response latency differences between nodes are significant, this value increases. The value is a non-negative integer. The larger the value, the higher the frequency of service execution failure and triggering compensation rollback within the current statistical period, and the worse the overall service availability. The count is incremented by one each time the node fails during orchestration and the system triggers a compensation rollback. This count is recorded regardless of whether the final compensation result is successful, pending, exhausted, or expired. ,because It reflects the frequency of compensation triggered by service failures, rather than the success or failure of the compensation itself. Take a non-negative value; when the average score of the overall resident evaluation decreases in the current statistical period compared to the previous statistical period, it indicates a negative result. Pick A positive value indicates a greater decline in resident satisfaction; a larger value indicates a larger decline in resident satisfaction. When the overall average resident evaluation score remains unchanged or increases during the statistical period... ,through After cutting Setting it to zero indicates that there are no abnormal signals in that dimension during the current period. This design makes the resident evaluation dimension sensitive only to a decline in satisfaction, and not to generating a reverse interference signal due to an increase in satisfaction.

[0054] This invention calculates the health score of service nodes based on three operational indicators: response latency, service availability, and resident evaluation, using a health assessment model. It then determines the weighting magnitude by combining the relative ranking factor within the group, ensuring that nodes with relatively better performance within the same group retain a reasonable share in scheduling. Simultaneously, it adaptively updates the weight coefficients of each dimension through an exponentially weighted moving average tracking volume and a dual-threshold bidirectional adjustment rule. This allows the health assessment model to automatically increase the sensitivity of corresponding dimensions when operational problems are prominent and automatically restore a balanced state after the problems subside, effectively suppressing the degradation problem of the assessment model caused by long-term unidirectional shifts in weight coefficients.

[0055] The above description is only 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 intelligent service system for a smart community, characterized in that, include: The service registration module is used to standardize and encapsulate multi-business heterogeneous services, including facility services and lifestyle services, using protocol conversion adapters and field mapping rules. The standard response latency benchmark value, group identifier, and initial allocation weight of each encapsulated service are written into the service registration directory to form a service resource pool. The request parsing module is used to perform structured parsing and service unit matching on resident service requests and governance decision event messages based on the service resource pool, so as to obtain standardized orchestration requests. The orchestration execution module is used to receive standardized orchestration requests, read the current scheduling policy, perform availability verification on the service nodes involved in the standardized orchestration requests, select the combined orchestration mode according to the dependency relationship type to drive the coordinated execution of each service node, record the call timestamp and execution status of each service node, count the actual response latency of each service node, and synchronize the operation data containing the actual response latency and execution status statistics of each service node to the scheduling optimization module. The scheduling optimization module is used to aggregate and calculate service node-level operational indicators according to statistical periods, determine the comprehensive health score of each service node through a health assessment model, generate a structured scheduling strategy based on the comprehensive health score and write it into the scheduling strategy storage, and adaptively update the weight coefficients of the health assessment model based on the exponentially weighted moving average tracking volume and the dual-threshold bidirectional adjustment rule.

2. The intelligent service system for a smart community as described in claim 1, characterized in that: The steps of the service registration module are as follows: For facility-type services, the communication protocol type of IoT devices is identified, the raw data streams reported by the devices are extracted, the raw data streams are encapsulated according to the standard event message structure, and the service quality level is assigned to them according to the urgency of the event data; the encapsulation result includes a unique stream identifier, timestamp sequence, device physical identifier, service event type field, real-time status field, and trigger command format specification; For lifestyle services, the original interface documents of the service providers are parsed. Based on the original interface documents, the interface protocol type, request parameter field list, response field list, and authentication method are extracted. A two-way mapping rule is established between the original fields and the standard fields of the intelligent service system, and the interface authentication method is configured. The two-way mapping rule converts the parameter fields of the standard calling protocol of the intelligent service system into personalized request parameters that conform to the service provider's interface definition according to the mapping rule in the request direction. In the response direction, the response fields returned by the service provider are converted into the standard response format of the intelligent service system. The conversion content covers parameter name mapping, data type conversion, and field merging processing. All business services are submitted for registration using a unified community service unit structure and written into the service registration directory.

3. The intelligent service system for a smart community as described in claim 1, characterized in that: The request parsing module includes a governance state machine and a parser. The specific steps of the request parsing module include: The governance state machine generates a globally unique deliberation identifier for each deliberation proposal. The process proceeds sequentially through five stages: solicitation of support, consultation and discussion, voting, public announcement of results, and follow-up on implementation, thereby obtaining information on governance decision-making events. After receiving the governance decision event message from the governance state machine, the parser encodes it according to the event type. To find the corresponding execution task template in the governance task template library for the retrieval key, extract field values ​​from the structured digest of the governance task template library according to the field mapping rules in the task template, and obtain the service unique identifier of the corresponding service unit in the service registry directory. and the API call address; among which the governance task template library is The primary key pre-stores a list of service type identifiers required for each type of meeting, the dependency types between service units, and a structured summary. The mapping rules from each field in the code to the service call parameters; The outputs of the governance state machine and the parser are encapsulated into a standardized orchestration request, including a request identifier. The system includes a trigger mode flag, an ordered set of service unit identifiers, a dependency type flag, and a request priority level. The request priority is carried in the request message or generated by mapping the urgency of governance decisions.

4. The intelligent service system for a smart community as described in claim 3, characterized in that: The governance state machine requires transitions between states. If the transition condition is met, the machine automatically advances to the next state; otherwise, it is set to the corresponding terminated state. The transition conditions are as follows: The conditions for transitioning from supporting the solicitation to consultation and discussion are as follows: ,in The number of supporters. Minimum number of support personnel required for operation and management; The conditions for transitioning from negotiation and discussion to voting are: ,in To determine the duration of the negotiations, The maximum negotiation time limit configured for the operations management party; The process of proceeding from the voting deadline to the results announcement must simultaneously meet the following conditions: the voting deadline has passed and... ,in, Indicates the approval rate of the votes in favor. This indicates the total number of valid affirmative votes. This represents the total number of valid votes. This indicates the threshold for the approval rate of proposals configured by the operations management team; When the voting deadline arrives The proposal state machine is set to the rejected and terminated state, and no governance decision event message is sent to the parser; Residents' real-name objections will be accepted starting from the result announcement period, and the cumulative number of valid objections during the announcement period will be counted. ,like If the state machine is set to the public withdrawal termination state and no governance decision event message is sent to the parser, then... The threshold for the proportion of objections triggered during the public notice period to be withdrawn is configured for the operation and management party; The conditions for transitioning from the result announcement state to the execution tracking state are: and Send a message carrying to the parser , Voting Results and Structured Summary News about governance decision-making events, among which The public notice period has been ongoing. To ensure the shortest possible public notice period; When the parser is If the search key does not find a corresponding template in the governance task template library, then Set the template configuration to a paused state and push it to the operations management team. Enumerated values, descriptive text of agenda item types, and A configuration request with already filled fields; After the operations management team completes the corresponding template configuration, Restore to the results public state, re-execute template retrieval and orchestration request encapsulation, add new template records to the governance task template library version management and increment the version number.

5. The intelligent service system for a smart community as described in claim 2, characterized in that: The specific steps of the orchestration execution module include: Read the scheduling policy version number from the scheduling policy store. The corresponding scheduling strategy, wherein the scheduling strategy records the service-unique identifier of each service node. The scheduling status and weight allocation are as follows: the scheduling status includes normal status, degraded status and offline status; the weight allocation is dynamically updated by the scheduling optimization module according to the health score, and the initial value is determined by the service registration module during registration according to the average score of the same group. Verify the scheduling status of each node in the ordered set of service unit identifiers in the standardized orchestration request: When a service node is in an offline state, and there is a service group identifier that is the same as that node... Among the other available nodes, the one with the highest current assigned weight is selected as the replacement node; When a service node is in a degraded scheduling state, its assigned weight is reduced. When a service node is in a normal scheduling state, it is executed in a queue according to the priority level of the request in the standardized orchestration request, where the priority is carried by the resident request message or generated by the urgency of the governance decision. The actual set of nodes involved in this execution and the version number of the scheduling strategy. Write the request identifier The corresponding execution logs are tagged with the orchestration method selected according to the dependency relationship type to drive the execution of each service node, and the call start timestamp is recorded for each service node. With return timestamp The actual response latency of the computing node : ； When each service node completes execution, the actual response latency and execution status are written to the execution log in real time. The execution status includes execution success, execution failure, timeout, compensation success, compensation business pending processing, compensation retry exhausted, or compensation timeout.

6. The intelligent service system for a smart community as described in claim 5, characterized in that: The orchestration methods include sequential orchestration, parallel orchestration, and conditional branch orchestration, wherein... The sequential orchestration method sends call requests to the interface call addresses of each service node in sequence, using the return result of the previous node as the input parameter for the next node. If a node exceeds a preset timeout period... If no response is received, its execution status is set to timeout failure. The parallel orchestration sends a call request to all service nodes marked as parallel orchestration in this request, and collects the return results of each node in an asynchronous waiting manner. After all nodes return successfully, the return results are summarized. If any node fails to execute, the execution fails and a compensation rollback is triggered. The conditional branch orchestration execution pre-service unit compares its return value with the preset branch condition set one by one, and enters the branch path that meets the condition to execute the corresponding sequential orchestration or parallel orchestration sub-process. The branch path that is not selected does not trigger any service unit call.

7. The intelligent service system for a smart community as described in claim 6, characterized in that: The compensation rollback specifically includes: Read the list of service nodes that have been successfully executed from the execution log, based on the compensation type flag of each node in the service registry. and compensation time window Calculate the remaining compensation period for each node in the list. : ； in The system moment when compensation is triggered. For the first Each service unit records the forward execution completion timestamp in the execution log. This represents an enumeration value of a completely reversible compensation type. An enumeration of compensation types that are irreversible but can be compensated by business operations; sort the list Compensation operations are triggered sequentially after sorting in ascending order: The service unit is directly marked as having exceeded the compensation period and a special alarm is pushed to the operation and management party; The node calls the compensation operation interface to undo the positive execution result. If the call fails, it retryes by increasing the retry interval exponentially in a backoff manner until the maximum number of retries is reached. If the process still fails, the execution status is marked as exhausted due to compensation retries. The node calls the compensation operation interface, records an exception flag, and initiates the business compensation process. Upon successful call, the node's execution status is updated to "compensation business pending" and written to the execution log. If the call fails, it is retried using an exponential backoff method. Second-rate.

8. The intelligent service system for a smart community as described in claim 1, characterized in that: The specific steps of the scheduling optimization module include: Statistical period As a time window, in each statistical period At the end, perform aggregate calculations on the operational data, based on the statistical period number. Identify each period, It is a positive integer; For the service registry entry Each service node, based on the actual response latency of all successful calls within this statistical period. The arithmetic mean of the actual average response delay is calculated. If no successful call occurs within this cycle, the current cycle will continue. ; Based on the number of successful calls within this period and number of failed calls Calculate the service availability rate ; The average resident evaluation score is calculated based on the arithmetic mean of the valid resident evaluation scores within this period. ; Will , , As input to the health assessment model, a comprehensive health score is calculated. ; Based on the comprehensive health score and the preset lower limit threshold of normal status and the lower limit threshold of the weighted state The relationships between these relationships determine the scheduling status of each service node, where : when The scheduling status is normal and the assigned weights remain unchanged; when When the scheduling state is reduced in weight, the reduction magnitude is based on the relative ranking factor within the group. Sure: ； in, For nodes Among all nodes in the same group identifier that are in normal or degraded status, sort by The ranking number when arranged from smallest to largest. This represents the total number of nodes in the same group identifier that are either in a normal or degraded state. when When the scheduling status is offline, the assigned weight is cleared to zero, and the remaining available nodes under the same group identifier perform weight normalization. The updated scheduling status and assigned weights are written into the scheduling policy storage, and the weight coefficients of the health assessment model are adaptively updated.

9. The intelligent service system for a smart community as described in claim 8, characterized in that: The formula for calculating the comprehensive health score is as follows: ； ； in For the first The service node is at the The overall health score for each statistical period. , , The response latency, service availability, and resident ratings were respectively measured in the [number]th month. Weighting coefficients for each statistical period The full score for resident evaluation; A piecewise mapping function for the response delay ratio, For response delay, high-quality gain coefficient, This represents the standard response latency baseline value preset during the registration of the i-th node.

10. The intelligent service system for a smart community as described in claim 9, characterized in that: The adaptive update of the weight coefficients of the health assessment model specifically includes: In each statistical period At the end, the original abnormal signal observations for the response latency dimension, service availability dimension, and resident evaluation were calculated respectively. ; For each dimension Maintain the index-weighted moving average tracking volume separately : ； in, It is the exponential smoothing coefficient. This represents the exponentially weighted moving average tracking value of the j-th dimension calculated and stored at the end of the (k-1)-th statistical period; according to With the enhanced trigger upper threshold and recovery trigger threshold Compare and calculate the weight adjustment increment for each dimension. : when hour, That is, for the first The weight coefficients for each dimension are increased by a fixed step size. ; when and hour, That is, for the first The weight coefficients of each dimension are reduced by a fixed step size. ; when The weights remain unchanged.

Images