feat: v2.0 — maintenance module, AI analysis, station power fix

- Add full 检修维护中心 (6.4): 3-type work orders (消缺/巡检/抄表),
  asset management, warehouse, work plans, billing settlement
- Add AI智能分析 tab with LLM-powered diagnostics (StepFun + ZhipuAI)
- Add AI模型配置 settings page (provider, temperature, prompts)
- Fix station power accuracy: use API station total (station_power)
  instead of inverter-level computation — eliminates timing gaps
- Add 7 new DB models, 4 new API routers, 5 new frontend pages
- Migrations: 009 (maintenance expansion) + 010 (AI analysis)
- Version bump: 1.6.1 → 2.0.0

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

core/backend/app/services/llm_service.py

@@ -0,0 +1,107 @@
+"""LLM service for AI-powered analysis using OpenAI-compatible APIs (StepFun, ZhipuAI)."""
+import logging
+from openai import AsyncOpenAI
+
+logger = logging.getLogger(__name__)
+
+
+async def get_llm_client(settings: dict, use_fallback: bool = False) -> tuple[AsyncOpenAI, str]:
+    """Create an OpenAI-compatible client based on settings.
+    Returns (client, model_name) tuple.
+    """
+    if use_fallback:
+        base_url = settings.get("ai_fallback_api_base_url", "")
+        api_key = settings.get("ai_fallback_api_key", "")
+        model = settings.get("ai_fallback_model_name", "codegeex-4")
+    else:
+        base_url = settings.get("ai_api_base_url", "")
+        api_key = settings.get("ai_api_key", "")
+        model = settings.get("ai_model_name", "step-2-16k")
+
+    client = AsyncOpenAI(base_url=base_url, api_key=api_key, timeout=30.0)
+    return client, model
+
+
+async def chat_completion(
+    messages: list[dict],
+    settings: dict,
+    temperature: float | None = None,
+    max_tokens: int | None = None,
+) -> str:
+    """Send a chat completion request with automatic fallback."""
+    temp = temperature or float(settings.get("ai_temperature", "0.7"))
+    tokens = max_tokens or int(settings.get("ai_max_tokens", "2000"))
+
+    # Try primary
+    try:
+        client, model = await get_llm_client(settings, use_fallback=False)
+        response = await client.chat.completions.create(
+            model=model,
+            messages=messages,
+            temperature=temp,
+            max_tokens=tokens,
+        )
+        return response.choices[0].message.content or ""
+    except Exception as e:
+        logger.warning(f"Primary LLM failed: {e}")
+
+        # Try fallback if enabled
+        if settings.get("ai_fallback_enabled") == "true":
+            try:
+                client, model = await get_llm_client(settings, use_fallback=True)
+                response = await client.chat.completions.create(
+                    model=model,
+                    messages=messages,
+                    temperature=temp,
+                    max_tokens=tokens,
+                )
+                return response.choices[0].message.content or ""
+            except Exception as e2:
+                logger.error(f"Fallback LLM also failed: {e2}")
+                raise Exception(f"All LLM providers failed. Primary: {e}, Fallback: {e2}")
+        else:
+            raise
+
+
+async def test_connection(settings: dict) -> dict:
+    """Test connection to the configured LLM provider."""
+    results = {"primary": {"status": "unknown"}, "fallback": {"status": "unknown"}}
+
+    # Test primary
+    try:
+        client, model = await get_llm_client(settings, use_fallback=False)
+        response = await client.chat.completions.create(
+            model=model,
+            messages=[{"role": "user", "content": "你好，请回复'连接成功'"}],
+            max_tokens=20,
+            temperature=0,
+        )
+        reply = response.choices[0].message.content or ""
+        results["primary"] = {
+            "status": "success",
+            "model": model,
+            "reply": reply[:100],
+        }
+    except Exception as e:
+        results["primary"] = {"status": "error", "error": str(e)[:200]}
+
+    # Test fallback
+    if settings.get("ai_fallback_enabled") == "true":
+        try:
+            client, model = await get_llm_client(settings, use_fallback=True)
+            response = await client.chat.completions.create(
+                model=model,
+                messages=[{"role": "user", "content": "你好，请回复'连接成功'"}],
+                max_tokens=20,
+                temperature=0,
+            )
+            reply = response.choices[0].message.content or ""
+            results["fallback"] = {
+                "status": "success",
+                "model": model,
+                "reply": reply[:100],
+            }
+        except Exception as e:
+            results["fallback"] = {"status": "error", "error": str(e)[:200]}
+
+    return results