Home Explore Help
Register Sign In
caijun
Repositories Public Activity

caijun pushed to dev at yanzheng/DualFlow

2 months ago

caijun pushed to dev at yanzheng/DualFlow

  • 4a63b728e4 fix: 补充全状态向量构建依赖模块

2 months ago

caijun pushed to dev at yanzheng/DualFlow

  • d8d9163050 refactor：删除非运行相关的绘图代码

2 months ago

caijun pushed to dev at yanzheng/DualFlow

  • 8bc4c4a7b7 fix：修改了轻量级决策脚本中的错误识别逻辑 refactor：优化了轻量级决策脚本中的决策状态生成逻辑

2 months ago

caijun pushed to dev at yanzheng/DualFlow

  • 9246590367 fix:复原了已有水厂的决策模型及调用参数文件

2 months ago

caijun pushed to dev at yanzheng/DualFlow

  • 04d56fb30c feat: 新增了结合短期预测模型的全状态决策模型接口

2 months ago

caijun pushed to dev at yanzheng/DualFlow

2 months ago

caijun pushed to dev at wangmingyang/DualFlow

  • 3f0532b1e6 feat:更新了带异常接口的超滤脚本，run_dqn_decide.py新增输入units_to_run

3 months ago

caijun pushed to dev at wangmingyang/DualFlow

  • f0c5ad1251 feat:1. 更新了带异常接口的反渗透脚本，当出现异常时，脚本返回调用脚本的时间及异常变量名 2. 更新了反渗透脚本的相关说明

3 months ago

caijun pushed to dev at wangmingyang/DualFlow

  • e8b1ee7532 Merge remote-tracking branch 'company/dev' into dev
  • fe9a3c902d feat:更新了反渗透模型的第一版代码，接口为models/ro_mechanism_predict/cip_predict/run_this.py，使用时需以下数据：1.指定机组RO1/2/3/4；2.进行cip时的压差上限 3.本周期已有的进水数据(本周期控制字为24.0阶段的数据)，数据格式见models/ro_mechanism_predict/use_data/test_cycle_sample.csv
  • ab03acd533 feat:增加了根据跨膜压差修正泵功率的代码
  • View comparison for these 3 commits »

3 months ago

caijun pushed to dev at wangmingyang/DualFlow

4 months ago

caijun pushed to dev at wangmingyang/DualFlow

  • b7f198fd21 Merge remote-tracking branch 'company/dev' into dev
  • 95230c0969 feat:1. 重新训练了锡山与龙亭的模型 2.优化了膜阻力相关的奖励计算，更新了兰考和安镇的模型
  • f989f54839 feat:1. 重新训练了锡山与龙亭的模型 2.优化了膜阻力相关的奖励计算，更新了兰考和安镇的模型
  • View comparison for these 3 commits »

4 months ago

caijun pushed to dev at wangmingyang/DualFlow

  • 517a240c86 Merge remote-tracking branch 'company/dev' into dev
  • eb67737d9c feat：增加了兰考和安镇的模型
  • 406baf4010 feat：增加了环境体中吨水电耗与药耗相关奖励的计算逻辑 debug：修复了数据读取文件中各传感器时间戳不一致导致合并数据过大的问题
  • d86a3e990d feat：增加了数据处理与物理模型中吨水电耗与药耗的计算逻辑
  • View comparison for these 4 commits »

4 months ago

caijun pushed to dev at wangmingyang/DualFlow

  • 7c4148008b Merge remote-tracking branch 'company/dev' into dev
  • 91bd4d50f8 debug：修复了简单调用模型时模型当前决策值超限直接报错的bug（本bug来源于训练时模型从3000-3800按60s步长取值，实际取值上限为3000+60*14=3840，后续训练吨水电耗/药耗的龙亭模型时一并修正）
  • View comparison for these 2 commits »

5 months ago

caijun pushed to dev at wangmingyang/DualFlow

5 months ago

caijun pushed to dev at wangmingyang/DualFlow

  • ca60aef51f 更新模型训练/调用架构说明md

5 months ago

caijun pushed to dev at wangmingyang/DualFlow

5 months ago

caijun pushed to dev at wangmingyang/DualFlow

  • fe768a3336 增加锡山水厂膜阻力48hCEB/48次CEB模型，相关模型及参数为models/uf-rl/xishan，调用入口为models/uf-rl/uf_train/rl_model/DQN/run_dqn_decide.py或models/uf-rl/uf_train/rl_model/DQN/run_dqn_deicde_totalstate.py
  • 1778ade1e0 增加锡山水厂膜阻力48hCEB/48次CEB模型，相关模型及参数为models/uf-rl/xishan，调用入口为models/uf-rl/uf_train/rl_model/DQN/run_dqn_decide.py或models/uf-rl/uf_train/rl_model/DQN/run_dqn_deicde_totalstate.py
  • 71a1b3d6bf Merge remote-tracking branch 'company/dev' into dev # Conflicts: # models/uf-rl/训练/uf_data_process/plot.py # models/uf-rl/训练/uf_train/env/reset_plot.py # models/uf-rl/训练/uf_train/rl_model/DQN/model/reward.png
  • 3dcc843ad2 2.10存档
  • View comparison for these 4 commits »

5 months ago

caijun pushed to dev at wangmingyang/DualFlow

  • e508df3f1c feat：真实工厂数据测试版调用代码 - 新增状态构建类DQNStateBuilder（models/uf-rl/uf_train/rl_model/DQN/dqn_statebuilder.py） - 新增全状态决策脚本（models/uf-rl/uf_train/rl_model/DQN/run_dqn_deicde_totalstate.py） - 示例文件位于models/uf-rl/uf_train/rl_model/DQN/online_datasets

6 months ago

caijun pushed to dev at wangmingyang/DualFlow

  • c694f02849 feat：真实工厂数据测试版调用代码 - 新增决策类UFDQNDecider（models/uf-rl/uf_train/rl_model/DQN/dqn_decider.py） - 新增决策类调用脚本run_dqn_decide.py - 决策模型位于models/uf-rl/uf_train/rl_model/DQN/model
  • eba5aa9521 feat：真实工厂数据测试版 - 优化渲染
  • View comparison for these 2 commits »

7 months ago