格致知微 EASON

格致知微｜负责人 EASON｜科研仿真优先｜MATLAB｜FDE·方案落地·AI应用｜多领域博士团队｜可交付项目

[格致知微 EASON @Easonyichen1]

「论文能跑，产线为何崩？」
格致知微博士团队可接：
① MATLAB/Simulink 科研仿真
② 储能 BMS·VPP·构网型
③ 企业级 FDE 与 AI 落地
海外线上｜国内先谈差旅+定金再驻场
（能力示意，非个案金额）

[格致知微 EASON @Easonyichen1]

@xpresstn Mandating storage with new solar >1 MW helps firming. Value shows when curtailment hours enter the model, not only GWh targets.

[格致知微 EASON @Easonyichen1]

@DPinyoloya Exactly—BESS can’t fix a weak offtaker/tariff. Model interconnection + SCR first; tech only works inside that envelope.

[格致知微 EASON @Easonyichen1]

@reyhanmerekar FDE scale = delete custom paths. Lock accept metric + change-sync owner before the first PR, or demos never leave the lab.

[格致知微 EASON @Easonyichen1]

@Kesse85585Chris Agree on edge vs polling. Treat runaway boundary as coupled with imbalance—hot cell under pulse precedes resting SOC delta.

[格致知微 EASON @Easonyichen1]

@olesen_buckstop Second-life packs need SOH under pulse, not rest OCV. Average kWh hides hot-cell imbalance that BMS flags late.

[格致知微 EASON @Easonyichen1]

@ChEngr67 FOPDT in MATLAB is fine for a first cut. Gate the fit with a known plant trip case—otherwise τ looks good and fails on step-up.

[格致知微 EASON @Easonyichen1]

@moiiikaaa Spot on. Best FDE signal = a failed accept metric they rewired ownership for—not a stack list.

[格致知微 EASON @Easonyichen1]

@sudheep8531 Agree—EPC has thin moat. Ask for cell-to-rack thermal coupling + SCR acceptance curves, not only MWh slides.

[格致知微 EASON @Easonyichen1]

@CES_Baker_Inst Grid-forming for data-center loads is the right frame. Weak SCR + large step → current limit hits before virtual inertia helps.

[格致知微 EASON @Easonyichen1]

@Materials_MRS Physics-informed SOH helps. Field trap we see: pulse-heat imbalance tagged as SOC drift when thermal coupling is ignored.

[格致知微 EASON @Easonyichen1]

@Saur_energy 10 GW/40 GWh + GFM PCS: lock SCR range and black-start timing in FAT—droop that passed SCR=5 often trips at SCR=2.

[格致知微 EASON @Easonyichen1]

@energystoragenw System security contracts for GFM BESS make sense. Without paid inertia/fault-ride windows, labs optimize MW, not post-fault authority.

[格致知微 EASON @Easonyichen1]

@SolarFred Hazelwood GFM PCS is the delivery cut. Acceptance usually fails on SCR sweep + current-limit timing, not nameplate MW.

[格致知微 EASON @Easonyichen1]

@SethChristx Portugal’s >45 MW GFM rule is the right KPI shift. Synthetic inertia ≠ SCR acceptance—current limit + post-fault damping still need an EMT freeze.

[格致知微 EASON @Easonyichen1]

@liaoyichen7 对齐。限流墙先撞时，H 再大也救不了虚拟阻抗抢权——验收只认 EMT 冻住的 fail 轨迹，不认 RMS 平滑。

[格致知微 EASON @Easonyichen1]

Same droop that passed SCR=5 can trip GFM BESS at SCR=2.

Lab gains fail on weak grids: lower SCR kills damping; half-step H error hits UVP on black-start.
Deliver: SCR sweep + damping + black-start timing (illus.).

Gezhi｜GFM BESS sims
Droop or inertia?

[格致知微 EASON @Easonyichen1]

短路比从5掉到2，同一套droop就能把构网储能抖到跳闸？

很多人把仿真调通的droop直接抄到弱电网现场。
SCR一降，小扰动阻尼塌掉；H再偏半档，黑启动会蹭低压保护。
交付：SCR扫描+阻尼图+黑启动时序（示意非实测）。

格致知微｜科研仿真·构网型储能
卡在droop还是惯量标定？

[格致知微 EASON @Easonyichen1]

@DrWangzhiqiang Europe's gap is industrial offtake, not lab talent. Added MW doesn't cut a factory energy-cost KPI until firming hours and interconnection sit in the model.

[格致知微 EASON @Easonyichen1]

@liaoyichen7 对齐。弱电网扫参我们也不把 H 加大当答案——电流限先撞墙时，收敛时间要按限流后的轨迹重算，不能用未限流曲线交差。

[格致知微 EASON @Easonyichen1]

@Longlonglongsi 在做。现场交付里我们把「验收数谁签 / 参数谁同步 / 出事谁在线」写成开工条件；缺一条宁可线上先对齐。你卡在哪一段？