iailab-plat.git

Merge remote-tracking branch 'origin/master'

Jay

2025-02-24 c3c7a6918f0e2dfe597c339117e4185b641be95f

提交 \| 用户 \| 时间
7fd198	1	package com.iailab.module.model.mdk.predict.impl;
潘	2
6957a3	3	import com.alibaba.fastjson.JSON;
7fd198	4	import com.alibaba.fastjson.JSONArray;
潘	5	import com.iail.model.IAILModel;
401492	6	import com.iailab.module.model.enums.CommonConstant;
373ab1	7	import com.iailab.module.model.common.enums.OutResultType;
1178da	8	import com.iailab.module.model.common.exception.ModelResultErrorException;
69bd5e	9	import com.iailab.module.model.mcs.pre.entity.MmItemOutputEntity;
7fd198	10	import com.iailab.module.model.mcs.pre.entity.MmModelArithSettingsEntity;
潘	11	import com.iailab.module.model.mcs.pre.entity.MmPredictModelEntity;
69bd5e	12	import com.iailab.module.model.mcs.pre.service.MmItemOutputService;
7fd198	13	import com.iailab.module.model.mcs.pre.service.MmModelArithSettingsService;
潘	14	import com.iailab.module.model.mdk.common.enums.TypeA;
	15	import com.iailab.module.model.mdk.common.exceptions.ModelInvokeException;
	16	import com.iailab.module.model.mdk.predict.PredictModelHandler;
	17	import com.iailab.module.model.mdk.sample.SampleConstructor;
	18	import com.iailab.module.model.mdk.sample.dto.SampleData;
	19	import com.iailab.module.model.mdk.vo.PredictResultVO;
45520a	20	import com.iailab.module.model.mpk.common.MdkConstant;
1a2b62	21	import com.iailab.module.model.mpk.common.utils.DllUtils;
7fd198	22	import lombok.extern.slf4j.Slf4j;
潘	23	import org.springframework.beans.factory.annotation.Autowired;
	24	import org.springframework.stereotype.Component;
	25
45520a	26	import java.util.Date;
D	27	import java.util.HashMap;
	28	import java.util.List;
	29	import java.util.Map;
7fd198	30
潘	31	/**
	32	* @author PanZhibao
	33	* @Description
	34	* @createTime 2024年09月01日
	35	*/
	36	@Slf4j
	37	@Component
	38	public class PredictModelHandlerImpl implements PredictModelHandler {
	39
	40	@Autowired
	41	private MmModelArithSettingsService mmModelArithSettingsService;
	42
	43	@Autowired
69bd5e	44	private MmItemOutputService mmItemOutputService;
7fd198	45
潘	46	@Autowired
	47	private SampleConstructor sampleConstructor;
	48
4f1717	49	/**
潘	50	* 根据模型预测,返回预测结果
	51	*
	52	* @param predictTime
	53	* @param predictModel
	54	* @return
	55	* @throws ModelInvokeException
	56	*/
7fd198	57	@Override
28c2db	58	public synchronized PredictResultVO predictByModel(Date predictTime, MmPredictModelEntity predictModel,String itemName,String itemNo) throws ModelInvokeException {
7fd198	59	PredictResultVO result = new PredictResultVO();
潘	60	if (predictModel == null) {
	61	throw new ModelInvokeException("modelEntity is null");
	62	}
	63	String modelId = predictModel.getId();
	64	try {
81ce77	65	List<SampleData> sampleDataList = sampleConstructor.constructSample(TypeA.Predict.name(), modelId, predictTime, itemName, new HashMap<>());
7fd198	66	String modelPath = predictModel.getModelpath();
潘	67	if (modelPath == null) {
373ab1	68	log.info("模型路径不存在，modelId=" + modelId);
7fd198	69	return null;
潘	70	}
	71	IAILModel newModelBean = composeNewModelBean(predictModel);
	72	HashMap<String, Object> settings = getPredictSettingsByModelId(modelId);
45520a	73	// 校验setting必须有pyFile，否则可能导致程序崩溃
D	74	if (!settings.containsKey(MdkConstant.PY_FILE_KEY)) {
	75	throw new RuntimeException("模型设置参数缺少必要信息【" + MdkConstant.PY_FILE_KEY + "】,请重新上传模型！");
	76	}
	77
7fd198	78	if (settings == null) {
潘	79	log.error("模型setting不存在，modelId=" + modelId);
	80	return null;
	81	}
	82	int portLength = sampleDataList.size();
	83	Object[] param2Values = new Object[portLength + 2];
	84	for (int i = 0; i < portLength; i++) {
4f1717	85	param2Values[i] = sampleDataList.get(i).getMatrix();
7fd198	86	}
潘	87	param2Values[portLength] = newModelBean.getDataMap().get("models");
4f1717	88	param2Values[portLength + 1] = settings;
7fd198	89
595ccf	90	log.info("####################### 预测模型 "+ "【itemId:" + predictModel.getItemid() + ",itemName：" + itemName + ",itemNo：" + itemNo + "】 ##########################");
b82ba2	91	// JSONObject jsonObjNewModelBean = new JSONObject();
D	92	// jsonObjNewModelBean.put("newModelBean", newModelBean);
	93	// log.info(String.valueOf(jsonObjNewModelBean));
	94	// JSONObject jsonObjParam2Values = new JSONObject();
	95	// jsonObjParam2Values.put("param2Values", param2Values);
6957a3	96	log.info("参数： " + JSON.toJSONString(param2Values));
7fd198	97
潘	98	//IAILMDK.run
1a2b62	99	HashMap<String, Object> modelResult = DllUtils.run(newModelBean, param2Values, predictModel.getMpkprojectid());
73a05d	100	//打印结果
D	101	log.info("预测模型计算完成：modelId=" + modelId + ",modelName=" + predictModel.getMethodname() + "，modelResult=" + JSON.toJSONString(modelResult));
	102	//判断模型结果
4f1717	103	if (!modelResult.containsKey(CommonConstant.MDK_STATUS_CODE) \|\| !modelResult.containsKey(CommonConstant.MDK_RESULT) \|\|
潘	104	!modelResult.get(CommonConstant.MDK_STATUS_CODE).toString().equals(CommonConstant.MDK_STATUS_100)) {
1178da	105	throw new ModelResultErrorException("模型结果异常：" + modelResult);
b2aca2	106	}
4f1717	107	modelResult = (HashMap<String, Object>) modelResult.get(CommonConstant.MDK_RESULT);
7fd198	108
373ab1	109	List<MmItemOutputEntity> itemOutputList = mmItemOutputService.getByItemid(predictModel.getItemid());
a6e46f	110	Map<MmItemOutputEntity, double[]> predictMatrixs = new HashMap<>();
373ab1	111	for (MmItemOutputEntity output : itemOutputList) {
潘	112	if (!modelResult.containsKey(output.getResultstr())) {
	113	continue;
	114	}
	115	OutResultType outResultType = OutResultType.getEumByCode(output.getResultType());
	116	switch (outResultType) {
	117	case D1:
	118	double[] temp1 = (double[]) modelResult.get(output.getResultstr());
	119	predictMatrixs.put(output, temp1);
	120	break;
	121	case D2:
	122	double[][] temp2 = (double[][]) modelResult.get(output.getResultstr());
	123	double[] tempColumn = new double[temp2.length];
	124	for (int i = 0; i < tempColumn.length; i++) {
	125	tempColumn[i] = temp2[i][output.getResultIndex()];
69bd5e	126	}
373ab1	127	predictMatrixs.put(output, tempColumn);
潘	128	break;
a6e46f	129	case D:
D	130	Double temp3 = (Double) modelResult.get(output.getResultstr());
51e472	131	predictMatrixs.put(output, new double[]{temp3});
a6e46f	132	break;
373ab1	133	default:
潘	134	break;
1a2b62	135	}
D	136	}
69bd5e	137	result.setPredictMatrixs(predictMatrixs);
1a2b62	138	result.setModelResult(modelResult);
7fd198	139	result.setPredictTime(predictTime);
50084d	140	} catch (ModelResultErrorException ex) {
efdc38	141	// ex.printStackTrace();
D	142	log.error("模型结果异常", ex);
50084d	143	throw ex;
D	144	} catch (Exception ex) {
c9dd12	145	// log.error("调用发生异常，异常信息为:{0}", ex.getMessage());
efdc38	146	// ex.printStackTrace();
D	147	log.error("模型运行异常", ex);
ead005	148	throw new ModelInvokeException(ex.getMessage());
7fd198	149	}
潘	150	return result;
	151	}
	152
	153	/**
fdcde1	154	* 预测，模拟调整
潘	155	*
	156	* @param predictTime
	157	* @param predictModel
	158	* @param itemName
	159	* @param itemNo
	160	* @return
	161	* @throws ModelInvokeException
	162	*/
	163	@Override
	164	public synchronized PredictResultVO predictByModel(Date predictTime, MmPredictModelEntity predictModel,String itemName,String itemNo, double[][] deviation) throws ModelInvokeException {
	165	PredictResultVO result = new PredictResultVO();
	166	if (predictModel == null) {
	167	throw new ModelInvokeException("modelEntity is null");
	168	}
	169	String modelId = predictModel.getId();
	170	try {
	171	List<SampleData> sampleDataList = sampleConstructor.constructSample(TypeA.Predict.name(), modelId, predictTime, itemName, new HashMap<>());
	172	String modelPath = predictModel.getModelpath();
	173	if (modelPath == null) {
	174	log.info("模型路径不存在，modelId=" + modelId);
	175	return null;
	176	}
	177	IAILModel newModelBean = composeNewModelBean(predictModel);
	178	HashMap<String, Object> settings = getPredictSettingsByModelId(modelId);
	179	// 校验setting必须有pyFile，否则可能导致程序崩溃
	180	if (!settings.containsKey(MdkConstant.PY_FILE_KEY)) {
	181	throw new RuntimeException("模型设置参数缺少必要信息【" + MdkConstant.PY_FILE_KEY + "】,请重新上传模型！");
	182	}
	183
	184	if (settings == null) {
	185	log.error("模型setting不存在，modelId=" + modelId);
	186	return null;
	187	}
	188	int portLength = sampleDataList.size();
	189	Object[] param2Values = new Object[portLength + 2];
	190	for (int i = 0; i < portLength; i++) {
	191	param2Values[i] = sampleDataList.get(i).getMatrix();
	192	}
	193	param2Values[portLength] = newModelBean.getDataMap().get("models");
	194	param2Values[portLength + 1] = settings;
	195
	196	log.info("####################### 模拟调整 "+ "【itemId:" + predictModel.getItemid() + ",itemName：" + itemName + ",itemNo：" + itemNo + "】 ##########################");
	197	log.info("参数： " + JSON.toJSONString(param2Values));
	198
	199	//IAILMDK.run
	200	HashMap<String, Object> modelResult = DllUtils.run(newModelBean, param2Values, predictModel.getMpkprojectid());
	201	//打印结果
	202	log.info("预测模型计算完成：modelId=" + modelId + ",modelName=" + predictModel.getMethodname() + "，modelResult=" + JSON.toJSONString(modelResult));
	203	//判断模型结果
	204	if (!modelResult.containsKey(CommonConstant.MDK_STATUS_CODE) \|\| !modelResult.containsKey(CommonConstant.MDK_RESULT) \|\|
	205	!modelResult.get(CommonConstant.MDK_STATUS_CODE).toString().equals(CommonConstant.MDK_STATUS_100)) {
	206	throw new ModelResultErrorException("模型结果异常：" + modelResult);
	207	}
	208	modelResult = (HashMap<String, Object>) modelResult.get(CommonConstant.MDK_RESULT);
	209
	210	List<MmItemOutputEntity> itemOutputList = mmItemOutputService.getByItemid(predictModel.getItemid());
	211	Map<MmItemOutputEntity, double[]> predictMatrixs = new HashMap<>();
	212	for (MmItemOutputEntity output : itemOutputList) {
	213	if (!modelResult.containsKey(output.getResultstr())) {
	214	continue;
	215	}
	216	OutResultType outResultType = OutResultType.getEumByCode(output.getResultType());
	217	switch (outResultType) {
	218	case D1:
	219	double[] temp1 = (double[]) modelResult.get(output.getResultstr());
	220	predictMatrixs.put(output, temp1);
	221	break;
	222	case D2:
	223	double[][] temp2 = (double[][]) modelResult.get(output.getResultstr());
	224	double[] tempColumn = new double[temp2.length];
	225	for (int i = 0; i < tempColumn.length; i++) {
	226	tempColumn[i] = temp2[i][output.getResultIndex()];
	227	}
	228	predictMatrixs.put(output, tempColumn);
	229	break;
	230	case D:
	231	Double temp3 = (Double) modelResult.get(output.getResultstr());
	232	predictMatrixs.put(output, new double[]{temp3});
	233	break;
	234	default:
	235	break;
	236	}
	237	}
	238	result.setPredictMatrixs(predictMatrixs);
	239	result.setModelResult(modelResult);
	240	result.setPredictTime(predictTime);
	241	} catch (ModelResultErrorException ex) {
	242	log.error("模型结果异常", ex);
	243	throw ex;
	244	} catch (Exception ex) {
	245	log.error("调用发生异常，异常信息为:{0}", ex.getMessage());
	246	throw new ModelInvokeException(ex.getMessage());
	247	}
	248	return result;
	249	}
	250	/**
7fd198	251	* 构造IAILMDK.run()方法的newModelBean参数
潘	252	*
	253	* @param predictModel
	254	* @return
	255	*/
	256	private IAILModel composeNewModelBean(MmPredictModelEntity predictModel) {
	257	IAILModel newModelBean = new IAILModel();
	258	newModelBean.setClassName(predictModel.getClassname().trim());
	259	newModelBean.setMethodName(predictModel.getMethodname().trim());
	260	//构造参数类型
	261	String[] paArStr = predictModel.getModelparamstructure().trim().split(",");
	262	Class<?>[] paramsArray = new Class[paArStr.length];
	263	for (int i = 0; i < paArStr.length; i++) {
	264	if ("[[D".equals(paArStr[i])) {
	265	paramsArray[i] = double[][].class;
	266	} else if ("Map".equals(paArStr[i]) \|\| "java.util.HashMap".equals(paArStr[i])) {
	267	paramsArray[i] = HashMap.class;
	268	}
	269	}
	270	newModelBean.setParamsArray(paramsArray);
	271	HashMap<String, Object> dataMap = new HashMap<>();
	272	HashMap<String, String> models = new HashMap<>(1);
b2aca2	273	models.put("model_path", predictModel.getModelpath());
7fd198	274	dataMap.put("models", models);
潘	275	newModelBean.setDataMap(dataMap);
	276	return newModelBean;
	277	}
	278
	279	/**
	280	* 根据模型id获取参数map
	281	*
	282	* @param modelId
	283	* @return
	284	*/
	285	private HashMap<String, Object> getPredictSettingsByModelId(String modelId) {
	286	List<MmModelArithSettingsEntity> list = mmModelArithSettingsService.getByModelId(modelId);
	287	HashMap<String, Object> result = new HashMap<>();
	288	for (MmModelArithSettingsEntity entry : list) {
	289	String valueType = entry.getValuetype().trim(); //去除两端空格
	290	if ("int".equals(valueType)) {
	291	int value = Integer.parseInt(entry.getValue());
	292	result.put(entry.getKey(), value);
	293	} else if ("double".equals(valueType)) {
	294	double value = Double.parseDouble(entry.getValue());
	295	result.put(entry.getKey(), value);
	296	} else if ("string".equals(valueType)) {
	297	String value = entry.getValue();
	298	result.put(entry.getKey(), value);
	299	} else if ("decimalArray".equals(valueType)) {
	300	JSONArray valueArray = JSONArray.parseArray(entry.getValue());
	301	double[] value = new double[valueArray.size()];
	302	for (int i = 0; i < valueArray.size(); i++) {
	303	value[i] = Double.parseDouble(valueArray.get(i).toString());
	304	}
	305	result.put(entry.getKey(), value);
	306	} else if ("decimal".equals(valueType)) {
	307	double value = Double.parseDouble(entry.getValue());
	308	result.put(entry.getKey(), value);
	309	}
	310	}
	311	return result;
	312	}
	313	}