|
| 1 | +""" |
| 2 | +src/core/fleet_orchestrator.py |
| 3 | +================================ |
| 4 | +BESSAI Edge Gateway — Multi-Site Fleet Orchestrator. |
| 5 | +
|
| 6 | +Manages telemetry, dispatch, and federated learning coordination across |
| 7 | +multiple BESSAI edge sites from a single orchestration process. |
| 8 | +
|
| 9 | +Architecture: |
| 10 | + - FleetOrchestrator is instantiated once per aggregation tier |
| 11 | + - Each SiteProxy represents a remote BESSAI edge gateway |
| 12 | + - Orchestrator polls sites, aggregates fleet KPIs, triggers VPP events |
| 13 | +
|
| 14 | +Typical deployment: |
| 15 | + - 1 fleet orchestrator per VPP control zone (e.g., per electricity market) |
| 16 | + - Each orchestrator coordinates N=5..50 BESSAI edge gateways |
| 17 | + - Communicates with edge via REST/gRPC (stub → mTLS in prod) |
| 18 | +
|
| 19 | +Usage:: |
| 20 | +
|
| 21 | + orch = FleetOrchestrator(vpp=VPPPublisher(), lca=LCAEngine()) |
| 22 | + orch.register_site("CL-001", SiteProxy("10.0.1.50", capacity_kwh=100)) |
| 23 | + orch.register_site("CL-002", SiteProxy("10.0.1.51", capacity_kwh=200)) |
| 24 | + summary = orch.run_cycle() |
| 25 | +""" |
| 26 | + |
| 27 | +from __future__ import annotations |
| 28 | + |
| 29 | +import asyncio |
| 30 | +import time |
| 31 | +from dataclasses import dataclass, field |
| 32 | +from typing import Optional |
| 33 | + |
| 34 | +import structlog |
| 35 | + |
| 36 | +from src.interfaces.metrics import ( |
| 37 | + FLEET_SITES_ACTIVE, |
| 38 | + FLEET_TOTAL_CAPACITY_KWH, |
| 39 | +) |
| 40 | + |
| 41 | +__all__ = [ |
| 42 | + "FleetOrchestrator", "SiteProxy", "SiteTelemetry", "FleetSummary", |
| 43 | +] |
| 44 | + |
| 45 | +log = structlog.get_logger(__name__) |
| 46 | + |
| 47 | + |
| 48 | +@dataclass |
| 49 | +class SiteTelemetry: |
| 50 | + """Telemetry snapshot from a single BESSAI edge site.""" |
| 51 | + site_id: str |
| 52 | + soc_pct: float # 0-100 |
| 53 | + power_kw: float # + = charging, - = discharging |
| 54 | + temp_c: float |
| 55 | + capacity_kwh: float |
| 56 | + available_kw: float |
| 57 | + anomaly_score: float = 0.0 |
| 58 | + timestamp: float = field(default_factory=time.time) |
| 59 | + |
| 60 | + |
| 61 | +@dataclass |
| 62 | +class FleetSummary: |
| 63 | + """Aggregated fleet KPIs for one orchestration cycle.""" |
| 64 | + n_sites: int |
| 65 | + total_capacity_kwh: float |
| 66 | + fleet_soc_pct: float # weighted average SOC |
| 67 | + total_power_kw: float # sum of all site power (kW) |
| 68 | + total_available_kw: float # sum of flex capacity |
| 69 | + sites_in_alarm: int # sites with anomaly_score > 0.7 |
| 70 | + cycle_duration_s: float |
| 71 | + timestamp: float = field(default_factory=time.time) |
| 72 | + |
| 73 | + def __repr__(self) -> str: |
| 74 | + return ( |
| 75 | + f"FleetSummary(n={self.n_sites}, " |
| 76 | + f"soc={self.fleet_soc_pct:.1f}%, " |
| 77 | + f"power={self.total_power_kw:.1f}kW, " |
| 78 | + f"flex={self.total_available_kw:.1f}kW)" |
| 79 | + ) |
| 80 | + |
| 81 | + |
| 82 | +class SiteProxy: |
| 83 | + """Proxy representing a remote BESSAI edge site. |
| 84 | +
|
| 85 | + In production this would use an async httpx/gRPC call. |
| 86 | + In simulation / testing, inject a ``telemetry_fn`` callback. |
| 87 | +
|
| 88 | + Parameters: |
| 89 | + host: IP/hostname of the remote site. |
| 90 | + site_id: Unique site identifier. |
| 91 | + capacity_kwh: Nameplate BESS capacity. |
| 92 | + telemetry_fn: Optional callable returning SiteTelemetry (for testing). |
| 93 | + """ |
| 94 | + |
| 95 | + def __init__( |
| 96 | + self, |
| 97 | + host: str, |
| 98 | + site_id: str = "unknown", |
| 99 | + capacity_kwh: float = 100.0, |
| 100 | + telemetry_fn=None, |
| 101 | + ) -> None: |
| 102 | + self.host = host |
| 103 | + self.site_id = site_id |
| 104 | + self.capacity_kwh = capacity_kwh |
| 105 | + self._telemetry_fn = telemetry_fn |
| 106 | + self._last_telemetry: Optional[SiteTelemetry] = None |
| 107 | + |
| 108 | + async def fetch_telemetry(self) -> SiteTelemetry: |
| 109 | + """Fetch current telemetry from this site. |
| 110 | +
|
| 111 | + Returns: |
| 112 | + SiteTelemetry — from inject fn (test), or a realistic stub. |
| 113 | + """ |
| 114 | + if self._telemetry_fn is not None: |
| 115 | + tel = self._telemetry_fn(self.site_id) |
| 116 | + self._last_telemetry = tel |
| 117 | + return tel |
| 118 | + |
| 119 | + # Production stub — would be: async with httpx.AsyncClient() as c: ... |
| 120 | + stub = SiteTelemetry( |
| 121 | + site_id=self.site_id, |
| 122 | + soc_pct=60.0, |
| 123 | + power_kw=0.0, |
| 124 | + temp_c=28.0, |
| 125 | + capacity_kwh=self.capacity_kwh, |
| 126 | + available_kw=self.capacity_kwh * 0.5, # 50% flex |
| 127 | + ) |
| 128 | + self._last_telemetry = stub |
| 129 | + return stub |
| 130 | + |
| 131 | + @property |
| 132 | + def last_telemetry(self) -> Optional[SiteTelemetry]: |
| 133 | + return self._last_telemetry |
| 134 | + |
| 135 | + |
| 136 | +class FleetOrchestrator: |
| 137 | + """Multi-site BESS fleet orchestrator. |
| 138 | +
|
| 139 | + Parameters: |
| 140 | + site_id: Identifier for Prometheus labels. |
| 141 | + anomaly_threshold: Score above which a site is counted as 'in alarm'. |
| 142 | + """ |
| 143 | + |
| 144 | + def __init__( |
| 145 | + self, |
| 146 | + site_id: str = "fleet", |
| 147 | + anomaly_threshold: float = 0.7, |
| 148 | + ) -> None: |
| 149 | + self.site_id = site_id |
| 150 | + self.anomaly_threshold = anomaly_threshold |
| 151 | + self._sites: dict[str, SiteProxy] = {} |
| 152 | + |
| 153 | + # ------------------------------------------------------------------ |
| 154 | + # Site management |
| 155 | + # ------------------------------------------------------------------ |
| 156 | + |
| 157 | + def register_site(self, site_id: str, proxy: SiteProxy) -> None: |
| 158 | + """Register a remote site under the given ID.""" |
| 159 | + self._sites[site_id] = proxy |
| 160 | + FLEET_SITES_ACTIVE.labels(site_id=self.site_id).set(len(self._sites)) |
| 161 | + log.info("fleet.site_registered", site_id=site_id, host=proxy.host) |
| 162 | + |
| 163 | + def remove_site(self, site_id: str) -> None: |
| 164 | + """Remove a site from the fleet.""" |
| 165 | + self._sites.pop(site_id, None) |
| 166 | + FLEET_SITES_ACTIVE.labels(site_id=self.site_id).set(len(self._sites)) |
| 167 | + log.info("fleet.site_removed", site_id=site_id) |
| 168 | + |
| 169 | + @property |
| 170 | + def n_sites(self) -> int: |
| 171 | + return len(self._sites) |
| 172 | + |
| 173 | + @property |
| 174 | + def total_capacity_kwh(self) -> float: |
| 175 | + return sum(p.capacity_kwh for p in self._sites.values()) |
| 176 | + |
| 177 | + # ------------------------------------------------------------------ |
| 178 | + # Orchestration cycle |
| 179 | + # ------------------------------------------------------------------ |
| 180 | + |
| 181 | + async def poll_all(self) -> list[SiteTelemetry]: |
| 182 | + """Poll all registered sites concurrently. |
| 183 | +
|
| 184 | + Returns: |
| 185 | + List of SiteTelemetry objects (one per site). |
| 186 | + """ |
| 187 | + coros = [proxy.fetch_telemetry() for proxy in self._sites.values()] |
| 188 | + results = await asyncio.gather(*coros, return_exceptions=True) |
| 189 | + telemetries = [] |
| 190 | + for result in results: |
| 191 | + if isinstance(result, Exception): |
| 192 | + log.error("fleet.poll_error", error=str(result)) |
| 193 | + else: |
| 194 | + telemetries.append(result) |
| 195 | + return telemetries |
| 196 | + |
| 197 | + def aggregate(self, telemetries: list[SiteTelemetry]) -> FleetSummary: |
| 198 | + """Aggregate telemetry list into a FleetSummary. |
| 199 | +
|
| 200 | + Args: |
| 201 | + telemetries: List of SiteTelemetry dataclasses. |
| 202 | +
|
| 203 | + Returns: |
| 204 | + FleetSummary KPIs. |
| 205 | + """ |
| 206 | + if not telemetries: |
| 207 | + return FleetSummary( |
| 208 | + n_sites=0, total_capacity_kwh=0.0, fleet_soc_pct=0.0, |
| 209 | + total_power_kw=0.0, total_available_kw=0.0, |
| 210 | + sites_in_alarm=0, cycle_duration_s=0.0, |
| 211 | + ) |
| 212 | + |
| 213 | + total_cap = sum(t.capacity_kwh for t in telemetries) |
| 214 | + weighted_soc = ( |
| 215 | + sum(t.soc_pct * t.capacity_kwh for t in telemetries) / total_cap |
| 216 | + if total_cap > 0 else 0.0 |
| 217 | + ) |
| 218 | + total_power = sum(t.power_kw for t in telemetries) |
| 219 | + total_avail = sum(t.available_kw for t in telemetries) |
| 220 | + alarms = sum(1 for t in telemetries if t.anomaly_score > self.anomaly_threshold) |
| 221 | + |
| 222 | + # Update Prometheus |
| 223 | + FLEET_TOTAL_CAPACITY_KWH.labels(site_id=self.site_id).set(total_cap) |
| 224 | + FLEET_SITES_ACTIVE.labels(site_id=self.site_id).set(len(telemetries)) |
| 225 | + |
| 226 | + return FleetSummary( |
| 227 | + n_sites=len(telemetries), |
| 228 | + total_capacity_kwh=total_cap, |
| 229 | + fleet_soc_pct=weighted_soc, |
| 230 | + total_power_kw=total_power, |
| 231 | + total_available_kw=total_avail, |
| 232 | + sites_in_alarm=alarms, |
| 233 | + cycle_duration_s=0.0, # populated by run_cycle() |
| 234 | + ) |
| 235 | + |
| 236 | + def run_cycle(self) -> FleetSummary: |
| 237 | + """Run a synchronous orchestration cycle (wraps async poll_all). |
| 238 | +
|
| 239 | + Returns: |
| 240 | + FleetSummary with aggregated fleet KPIs. |
| 241 | + """ |
| 242 | + t0 = time.perf_counter() |
| 243 | + try: |
| 244 | + loop = asyncio.get_event_loop() |
| 245 | + except RuntimeError: |
| 246 | + loop = asyncio.new_event_loop() |
| 247 | + asyncio.set_event_loop(loop) |
| 248 | + |
| 249 | + telemetries = loop.run_until_complete(self.poll_all()) |
| 250 | + summary = self.aggregate(telemetries) |
| 251 | + summary.cycle_duration_s = time.perf_counter() - t0 |
| 252 | + |
| 253 | + log.info( |
| 254 | + "fleet.cycle_complete", |
| 255 | + n_sites=summary.n_sites, |
| 256 | + fleet_soc=round(summary.fleet_soc_pct, 1), |
| 257 | + total_power_kw=round(summary.total_power_kw, 1), |
| 258 | + alarms=summary.sites_in_alarm, |
| 259 | + duration_s=round(summary.cycle_duration_s, 4), |
| 260 | + ) |
| 261 | + return summary |
0 commit comments