feat(alert): 新增 ANN 模型训练与测试功能

- 添加 ANN 模型训练接口 /alert/model/train/ann
- 添加 ANN 模型测试接口 /alert/model/test/ann
- 扩展 ModelInfo 和 ModelInitVO 支持 ANN 层级和迭代参数
- 更新模型服务接口和实现以支持 ANN 特定逻辑
- 重构训练参数结构以区分 PCA 与 ANN 配置
- 修复模型创建时算法相关字段设置问题
- 增强模型更新时对不同算法参数的处理逻辑
- 优化点数据结构增加 type 字段支持

yudao-module-alert/yudao-module-alert-biz/src/main/java/cn/iocoder/yudao/module/alert/controller/admin/model/ModelController.java

@@ -2,6 +2,8 @@ package cn.iocoder.yudao.module.alert.controller.admin.model;
 
 import cn.iocoder.yudao.framework.common.pojo.CommonResult;
 import cn.iocoder.yudao.module.alert.controller.admin.model.vo.*;
+import cn.iocoder.yudao.module.alert.param.AnnTestParam;
+import cn.iocoder.yudao.module.alert.param.AnnTrainParam;
 import cn.iocoder.yudao.module.alert.param.ModelTestParam;
 import cn.iocoder.yudao.module.alert.param.TrainParam;
 import cn.iocoder.yudao.module.alert.service.model.ModelService;
@@ -72,12 +74,28 @@ public class ModelController {
         }
     }
 
+    @PostMapping("/train/ann")
+    public CommonResult<String> getAnnTrainData(@RequestBody AnnTrainParam param) {
+        try {
+            String trainInfo = modelService.trainAnn(param);
+            return CommonResult.success(trainInfo);
+        } catch (Exception e) {
+            return CommonResult.error(INTERNAL_SERVER_ERROR.getCode(), e.getMessage());
+        }
+    }
+
     @PostMapping("/test")
     public CommonResult<ModelTestData> getTestData(@RequestBody ModelTestParam param) {
         ModelTestData modelTestData = modelService.getModelTestData(param);
         return CommonResult.success(modelTestData);
     }
 
+    @PostMapping("/test/ann")
+    public CommonResult<ModelTestData> getAnnTestData(@RequestBody AnnTestParam param) {
+        ModelTestData modelTestData = modelService.getAnnModelTestData(param);
+        return CommonResult.success(modelTestData);
+    }
+
     /**
      * 创建草稿版本（v-test）
      */

yudao-module-alert/yudao-module-alert-biz/src/main/java/cn/iocoder/yudao/module/alert/controller/admin/model/model/ModelInfo.java

@@ -69,6 +69,12 @@ public class ModelInfo {
     @JsonProperty("rate")
     private String rate;
 
+    @JsonProperty("layer")
+    private String layer;
+
+    @JsonProperty("iter")
+    private Integer iter;
+
     @JsonProperty("outPointInfo")
     private List<Point> outPointInfo;
 
@@ -102,7 +108,7 @@ public class ModelInfo {
     private AlarmModelSet alarmModelSet;
 
     @JsonProperty("para")
-    private TrainInfo para;
+    private Object para;
 
     @JsonProperty("principal")
     private Integer principal;

yudao-module-alert/yudao-module-alert-biz/src/main/java/cn/iocoder/yudao/module/alert/controller/admin/model/model/Point.java

@@ -57,4 +57,6 @@ public class Point {
     private BigDecimal C95;
     @JsonProperty("C99")
     private BigDecimal C99;
+
+    private boolean type;
 }

yudao-module-alert/yudao-module-alert-biz/src/main/java/cn/iocoder/yudao/module/alert/controller/admin/model/vo/ModelInitVO.java

@@ -57,4 +57,14 @@ public class ModelInitVO {
     private Integer sampling;
 
     private String rate;
+
+    /**
+     * ANN 模型层级配置
+     */
+    private String layer;
+
+    /**
+     * ANN 迭代次数
+     */
+    private Integer iter;
 }

yudao-module-alert/yudao-module-alert-biz/src/main/java/cn/iocoder/yudao/module/alert/param/TrainParam.java

@@ -16,11 +16,36 @@ public class TrainParam {
     @JsonProperty("Train_Data")
     private TrainData trainData;
 
+    /**
+     * 算法类型：PCA / ANN 等
+     */
+    private String type;
+
+    /**
+     * 条件（保留原拼写以兼容旧入参）
+     */
+    private String conditon;
+
+    /**
+     * 正确拼写的条件字段，兼容新入参
+     */
+    @JsonAlias("condition")
+    private String condition;
+
+    /**
+     * PCA 专用配置
+     */
+    private PcaParam pca;
+
+    /**
+     * ANN 专用配置（如仍走旧接口可使用）
+     */
+    private AnnParam ann;
+
+    // === 旧字段兼容 ===
     @JsonAlias("Hyper_para")
     @JsonProperty("Hyper_para")
     private HyperPara hyperPara;
-    private String type;
-    private String conditon;
 
     @JsonAlias("smote_config")
     @JsonProperty("smote_config")
@@ -30,6 +55,10 @@ public class TrainParam {
 
     private String targetPoint;
 
+    private String iter;
+
+    private String hide;
+
 
     @Data
     public static class Smote {
@@ -57,4 +86,29 @@ public class TrainParam {
         private String uplow;
         private Integer interval;
     }
+
+    @Data
+    @AllArgsConstructor
+    @NoArgsConstructor
+    public static class PcaParam {
+        @JsonAlias("Hyper_para")
+        @JsonProperty("Hyper_para")
+        private HyperPara hyperPara;
+
+        @JsonAlias("smote_config")
+        @JsonProperty("smote_config")
+        private List<Object> smoteConfig;
+
+        private Boolean smote;
+
+        private String targetPoint;
+    }
+
+    @Data
+    @AllArgsConstructor
+    @NoArgsConstructor
+    public static class AnnParam {
+        private String iter;
+        private String hide;
+    }
 }

yudao-module-alert/yudao-module-alert-biz/src/main/java/cn/iocoder/yudao/module/alert/service/model/ModelService.java

@@ -2,7 +2,9 @@ package cn.iocoder.yudao.module.alert.service.model;
 
 
 import cn.iocoder.yudao.module.alert.controller.admin.model.vo.*;
+import cn.iocoder.yudao.module.alert.param.AnnTestParam;
 import cn.iocoder.yudao.module.alert.param.ModelTestParam;
+import cn.iocoder.yudao.module.alert.param.AnnTrainParam;
 import cn.iocoder.yudao.module.alert.param.TrainParam;
 
 import java.util.List;
@@ -55,8 +57,12 @@ public interface ModelService {
 
     TrainInfo trainModel(TrainParam param);
 
+    String trainAnn(AnnTrainParam param);
+
     ModelTestData getModelTestData(ModelTestParam param);
 
+    ModelTestData getAnnModelTestData(AnnTestParam param);
+
     /**
      * 模型下装：校验训练/评估后，将模型状态置为已下装并落库版本
      *

yudao-module-alert/yudao-module-alert-biz/src/main/java/cn/iocoder/yudao/module/alert/service/model/impl/ModelServiceImpl.java

@@ -20,6 +20,8 @@ import cn.iocoder.yudao.module.alert.dao.mapper.AssessReportCfgMapper;
 import cn.iocoder.yudao.module.alert.dao.service.ModelCfgService;
 import cn.iocoder.yudao.module.alert.dao.service.ModelVersionService;
 import cn.iocoder.yudao.module.alert.dao.service.SystemCfgService;
+import cn.iocoder.yudao.module.alert.param.AnnTestParam;
+import cn.iocoder.yudao.module.alert.param.AnnTrainParam;
 import cn.iocoder.yudao.module.alert.param.ModelTestParam;
 import cn.iocoder.yudao.module.alert.param.TrainParam;
 import cn.iocoder.yudao.module.alert.service.model.ModelService;
@@ -124,11 +126,12 @@ public class ModelServiceImpl implements ModelService {
     @Override
     public Integer createModel(ModelInitVO model) {
         try {
+            Algorithm algorithm = model.getAlgorithm();
             ModelInfo info = new ModelInfo();
             BeanUtil.copyProperties(model, info);
             ModelCfg modelCfg = ModelCfg.builder()
                     .systemId(model.getSystemId())
-                    .algorithmId(model.getAlgorithm().code)
+                    .algorithmId(algorithm.code)
                     .modelName(model.getModelName())
                     .createTime(new Date())
                     .trash(ModelTrash.NORMAL.code)
@@ -141,13 +144,25 @@ public class ModelServiceImpl implements ModelService {
             info.setId(String.valueOf(modelId));
             info.setSystemId(model.getSystemId());
             info.setUnit(model.getUnit());
-            info.setAlgorithm(model.getAlgorithm());
+            info.setAlgorithm(algorithm);
             info.setFounder(modelCfg.getCreator());
             info.setCreateTime(modelCfg.getCreateTime());
             info.setAlarmModelSet(ModelInfo.AlarmModelSet.defaultInit());
             info.setName(model.getModelName());
             info.setDescription(model.getDescription());
-            info.setRate(model.getRate());
+            if (Algorithm.PCA.equals(algorithm)) {
+                info.setRate(model.getRate());
+                info.setLayer(null);
+                info.setIter(null);
+            } else if (Algorithm.ANN.equals(algorithm)) {
+                info.setLayer(model.getLayer());
+                info.setRate(null);
+                info.setIter(model.getIter());
+            } else {
+                info.setRate(model.getRate());
+                info.setLayer(model.getLayer());
+                info.setIter(model.getIter());
+            }
             info.setTrainTime(new ArrayList<>());
             modelCfg.setModelInfo(JsonUtils.toJsonString(info));
             modelCfgService.updateById(modelCfg);
@@ -161,6 +176,19 @@ public class ModelServiceImpl implements ModelService {
 
     @Override
     public Boolean updateModelInfo(ModelInfoVO modelInfo) {
+        ModelCfg existCfg = modelCfgService.getById(Integer.parseInt(modelInfo.getId()));
+        if (existCfg != null) {
+            Algorithm algorithm = Algorithm.of(existCfg.getAlgorithmId());
+            modelInfo.setAlgorithm(algorithm);
+            if (Algorithm.PCA.equals(algorithm)) {
+                modelInfo.setLayer(null);
+                modelInfo.setIter(null);
+            } else if (Algorithm.ANN.equals(algorithm)) {
+                modelInfo.setRate(null);
+            } else {
+                // 其他算法默认保留传入字段
+            }
+        }
         modelInfo.setModifier(SecurityFrameworkUtils.getLoginUserNickname());
         modelInfo.setModifiedTime(new Date());
         ModelCfg modelCfg = ModelCfg.builder()
@@ -196,7 +224,29 @@ public class ModelServiceImpl implements ModelService {
 
     @Override
     public TrainInfo trainModel(TrainParam param) {
-        String trainBody = HttpUtils.post(algorithmHost + "/api/test/ClearTrain", null, JsonUtils.toJsonString(param));
+        if ("ANN".equalsIgnoreCase(param.getType())) {
+            throw new RuntimeException("请使用 /alert/model/train/ann 接口进行 ANN 训练");
+        }
+        return trainPca(param);
+    }
+
+    private TrainInfo trainPca(TrainParam param) {
+        Map<String, Object> payload = new HashMap<>();
+        payload.put("Train_Data", param.getTrainData());
+        payload.put("type", param.getType());
+        payload.put("condition", param.getCondition());
+        payload.put("conditon", param.getConditon());
+
+        TrainParam.PcaParam pca = param.getPca();
+        TrainParam.HyperPara hyperPara = pca != null && pca.getHyperPara() != null ? pca.getHyperPara() : param.getHyperPara();
+        if (hyperPara != null) {
+            payload.put("Hyper_para", hyperPara);
+        }
+        payload.put("smote_config", pca != null ? pca.getSmoteConfig() : param.getSmoteConfig());
+        payload.put("smote", pca != null && pca.getSmote() != null ? pca.getSmote() : param.getSmote());
+        payload.put("targetPoint", pca != null ? pca.getTargetPoint() : param.getTargetPoint());
+
+        String trainBody = HttpUtils.post(algorithmHost + "/api/test/ClearTrain", null, JsonUtils.toJsonString(payload));
         if (trainBody.contains("error_msg")) {
             throw new RuntimeException("模型训练异常：" + JsonUtils.parseObject(trainBody,
                     new TypeReference<Map<String, String>>() {
@@ -206,6 +256,42 @@ public class ModelServiceImpl implements ModelService {
         });
     }
 
+    @Override
+    public String trainAnn(AnnTrainParam param) {
+        Map<String, Object> payload = new HashMap<>();
+        TrainParam.TrainData trainData = param.getTrainData();
+        String time = trainData != null ? trainData.getTime() : null;
+        payload.put("time", time == null ? null : time.replaceAll(";+$", ""));
+        payload.put("point", trainData != null ? trainData.getPoints() : null);
+        Integer interval = trainData != null ? trainData.getInterval() : null;
+        payload.put("interval", interval == null ? null : interval * 1000);
+
+        String iter = param.getIter();
+        String hide = param.getHide();
+
+        payload.put("iter", iter);
+        payload.put("layer", StringUtils.hasText(hide) ? hide.split("-") : new String[]{});
+        payload.put("type", param.getType());
+        payload.put("dead", trainData != null ? trainData.getDead() : null);
+        payload.put("limit", trainData != null ? trainData.getLimit() : null);
+        payload.put("uplow", trainData != null ? trainData.getUplow() : null);
+        payload.put("condition", resolveCondition(param));
+        String trainBody = HttpUtils.post(algorithmHost + "/api/test/ANN_Train", null, JsonUtils.toJsonString(payload));
+        if (trainBody.contains("error_msg")) {
+            throw new RuntimeException("模型训练异常：" + JsonUtils.parseObject(trainBody,
+                    new TypeReference<Map<String, String>>() {
+                    }).get("error_msg"));
+        }
+        return trainBody;
+    }
+
+    private String resolveCondition(AnnTrainParam param) {
+        if (StringUtils.hasText(param.getCondition())) {
+            return param.getCondition();
+        }
+        return param.getConditon();
+    }
+
 
     @Override
     public ModelTestData getModelTestData(ModelTestParam param) {
@@ -214,6 +300,21 @@ public class ModelServiceImpl implements ModelService {
         return JsonUtils.parseObject(result, ModelTestData.class);
     }
 
+    @Override
+    public ModelTestData getAnnModelTestData(AnnTestParam param) {
+        Map<String, Object> payload = new HashMap<>();
+        payload.put("point", param.getPoints());
+        payload.put("model", param.getModel());
+        payload.put("time", param.getTime());
+        Integer interval = param.getInterval();
+        payload.put("interval", interval == null ? null : interval);
+        payload.put("type", param.getType());
+
+        String result = HttpUtils.post(algorithmHost + "/api/test/ANN_Test", null, JsonUtils.toJsonString(payload))
+                .replace("NaN", "-1").replace("Infinity", "1");
+        return JsonUtils.parseObject(result, ModelTestData.class);
+    }
+
     @Override
     @Transactional(rollbackFor = Exception.class)
     public ModelInfoVO bottomModel(Integer id, Long reportId) {