Operational data accumulated by the crane's predictive health management (PHM) system only delivers its full value when integrated with the workshop's manufacturing execution system (MES) and higher-level management platforms — transforming it from an isolated alerting tool into a data hub embedded in production scheduling decisions. Kelude has now officially opened the RESTful API of its overhead crane PHM predictive maintenance system, with data interfacing and joint commissioning completed with 30 MES vendors as of press time, covering three major system categories: metallurgical MES, discrete manufacturing MES, and warehouse WMS. With the API now open, real-time crane status data, alert events, and maintenance recommendations can be pushed directly to customers' existing production management platforms — eliminating the need for operators to switch between multiple systems. This article provides a systematic overview across four dimensions: API technical specifications, integration workflow, typical application scenarios, and commissioning results.

PHM API Architecture and Data Model

The PHM API follows the OpenAPI 3.0 specification and adopts a RESTful architecture, supporting both JSON and Protocol Buffers for data serialization. The API is organized into five resource categories, covering the full core data capabilities of the crane PHM system.

Realtime Status API: Provides one-click query of the crane's current operating status. Key endpoints include: GET /api/v1/crane/{craneId}/status (returns current operating parameters — crane bridge X/Y coordinates, lifting height, travel speed, motor current, brake status, overload limiter readings, and 36 fields in total, refreshed every 1 second), GET /api/v1/crane/{craneId}/health (returns the equipment health score, a 0–100 health index computed from multi-dimensional parameter fusion, updated every 5 minutes), and GET /api/v1/crane/{craneId}/alerts (returns the current list of unresolved alerts, filterable by severity level and alert type).

History API: Provides batch query of operational data and alert history. Endpoints include: GET /api/v1/crane/{craneId}/timeseries (queries historical time-series data for specified parameters over a time range, supporting continuous data export of up to 30 days, with sampling intervals configurable at 1 second / 10 seconds / 1 minute), GET /api/v1/crane/{craneId}/alerts/history (queries historical alert event lists, including alert generation time, clearing time, alert type, severity level, and handling status), and GET /api/v1/crane/{craneId}/maintenance (queries equipment maintenance records, including maintenance type, time, content, and executor). The History API is primarily used by MES systems to generate equipment OEE analysis reports and maintenance plan schedules.

Alert Webhook API: Enables MES platforms to subscribe to PHM alert events via webhooks. Customers configure alert push rules in the PHM web console (specifying the push URL, alert severity threshold, and push frequency). When the PHM system generates a new alert event, it automatically sends a POST request to the configured webhook URL with the alert event data in JSON format. The webhook supports a retry mechanism — if the first push fails, it retries after 30 seconds, up to 3 attempts total; if all three attempts fail, it falls back to email notification for manual handling by an administrator. The median webhook push latency is 5 seconds (from PHM anomaly detection to receipt of the push message by the customer's MES).

Device Management API: Provides CRUD operations for basic crane equipment information. Endpoints support adding new equipment (POST /api/v1/crane, which auto-generates a unique device identifier and initial configuration parameters), updating equipment configuration (PATCH /api/v1/crane/{craneId}, for modifying alert thresholds, sampling frequency, and other parameters), querying the equipment list (GET /api/v1/cranes, filterable by workshop, equipment type, and operating status), and equipment deletion (DELETE /api/v1/crane/{craneId}, requiring secondary confirmation). The Device Management API allows customers' MES systems to automatically complete PHM device registration when adding a new crane, eliminating the need for manual login to the PHM console.

Model API: Provides AI diagnostic model management and diagnostic result queries. Endpoints include: GET /api/v1/model/{modelId}/info (queries information about a specified diagnostic model — model version number, deployment time, supported crane models, and accuracy indicators), POST /api/v1/crane/{craneId}/diagnose (triggers an immediate health diagnosis for a specified crane, returning the health status of each subsystem and recommended maintenance actions), and GET /api/v1/crane/{craneId}/diagnose/history (queries the list of historical diagnostic reports). The Model API enables customers' MES dispatching systems to proactively check crane health status before production scheduling, avoiding assignment of production tasks to equipment in poor health.

API Integration Technical Solution and Security Certification

The PHM API uses OAuth 2.0 Client Credentials Grant for authentication. During initial integration, the customer's MES system applies for a client ID and client secret through the PHM developer platform, then obtains an access token via the POST /oauth/token endpoint. The token is valid for 24 hours by default and must be refreshed using the refresh_token after expiration. API request rate limits are as follows: the Realtime Status API allows up to 6,000 queries per hour (approximately 1.67 per second on average), the History API allows up to 360 queries per hour, and the Alert Webhook and Model APIs have no frequency limits. API response time is under 500 ms at the 99th percentile (excluding slow network transmission).

Data transmission is encrypted with TLS 1.3, and the API gateway is deployed on Alibaba Cloud API Gateway, with support for WAF (Web Application Firewall) and DDoS protection. API request logs record the timestamp, client ID, request path, response status code, and response duration for all calls, with a log retention period of 180 days. Customers can self-service query API call statistics (daily call volume, success rate, average response time, and other indicators) through the developer platform. The platform provides an API sandbox environment where customers can develop and debug interfaces using simulated data before switching to the production environment. The sandbox environment is fully isolated from production data and does not affect real equipment data.

Integration Progress and MES Vendor Coverage

As of press time, API joint commissioning has been completed with 30 MES vendors, covering four industry segments: metallurgy (12 vendors), heavy equipment manufacturing (8 vendors), machining (6 vendors), and warehousing and logistics (4 vendors). The integrated MES system types include: mainstream domestic MES platforms (such as Yonyou U8+MES, Kingdee Cloud Galaxy MES, Digiwin MES, etc. — 14 vendors), vertical-industry MES platforms (such as metallurgical MES, aluminum processing MES, shipbuilding MES, etc. — 10 vendors), and self-developed MES systems (workshop management systems developed in-house by large enterprises — 6 vendors).

During integration, MES vendors typically use three common patterns to incorporate PHM data: Alert dashboard integration (embedding the PHM alert list directly into the MES equipment monitoring dashboard so operators can view overhead crane health alerts without switching systems — chosen by approximately 70% of customers), Automated maintenance work order generation (the MES system receives PHM alerts and automatically triggers the work order creation process, pre-populating equipment information, alert descriptions, and recommended maintenance plans — chosen by approximately 45% of customers), and Equipment health–linked scheduling (the MES scheduling module queries the PHM equipment health score when assigning lifting tasks, prioritizing urgent jobs to cranes with higher health scores — chosen by approximately 20% of customers). These three patterns can be combined and do not conflict with one another.

The average integration testing cycle is 3–5 business days (from receiving the MES vendor's API documentation to completing end-to-end data validation), with the fastest vendor completing integration in just 2 business days. Typical issues identified during joint testing include: API request rate limits being exceeded (resolved by adjusting the MES polling interval — a polling period of no less than 5 seconds is recommended for real-time status endpoints), data synchronization timing issues (timezone mismatches between PHM alert timestamps and MES work order times — resolved by returning both UTC and local time in alert payloads), and expired Webhook receiver certificates (resolved through automated certificate validity monitoring via the API Gateway with 30-day advance renewal alerts).

Typical Application Scenarios and Performance Results

Take a metallurgical enterprise with 5 overhead cranes that has integrated the PHM API. Before integration, equipment administrators had to log into the PHM Web backend to manually check for crane alerts, then fill out paper maintenance work orders and deliver them to the maintenance crew. The average time from alert generation to work order dispatch was 45 minutes. After API integration, PHM alert events are automatically pushed to the enterprise's MES system via Webhook. The MES system then generates maintenance work orders automatically based on alert severity — Warning-level alerts are routed to the day-shift maintenance team, while Critical-level alerts immediately notify the on-duty engineer. Work order generation time has been reduced to under 10 seconds. Six months of post-integration data show that the average response time for unplanned crane downtime (from fault occurrence to maintenance personnel arrival on site) dropped from 85 minutes to 22 minutes, and production capacity losses caused by unplanned downtime were reduced by approximately 62%.

The open API interface also provides a valuable data foundation for continuous PHM system optimization. The crane health score early warning model iteratively trains using maintenance result data fed back from the MES system — including whether the fault was indeed predicted by the PHM alert, the actual replacement component models, and maintenance duration — improving early warning accuracy by approximately 1.5 percentage points per quarter. Data interfaces planned for the next phase include: crane energy consumption data (real-time and cumulative power draw for the crane bridge, trolley, and hoisting motors, supporting energy benchmarking and efficiency optimization) and lifting operation statistics (daily lift counts per crane, load weight distribution, and operational duration metrics to support production scheduling efficiency analysis). The goal is to complete integration with 50 MES vendors by year-end.

FAQ