iailab-module-model/iailab-module-model-biz/src/main/java/com/iailab/module/model/mdk/predict/PredictModelHandler.java
@@ -23,4 +23,16 @@ * @throws ModelInvokeException */ PredictResultVO predictByModel(Date predictTime, MmPredictModelEntity predictModel,String itemName,String itemNo) throws ModelInvokeException; /** * 预测，模拟调整 * * @param predictTime * @param predictModel * @param itemName * @param itemNo * @param deviation * @return */ PredictResultVO predictByModel(Date predictTime, MmPredictModelEntity predictModel,String itemName,String itemNo, double[][] deviation) throws ModelInvokeException; } iailab-module-model/iailab-module-model-biz/src/main/java/com/iailab/module/model/mdk/predict/impl/PredictModelHandlerImpl.java
@@ -151,6 +151,103 @@ } /** * 预测，模拟调整 * * @param predictTime * @param predictModel * @param itemName * @param itemNo * @return * @throws ModelInvokeException */ @Override public synchronized PredictResultVO predictByModel(Date predictTime, MmPredictModelEntity predictModel,String itemName,String itemNo, double[][] deviation) throws ModelInvokeException { PredictResultVO result = new PredictResultVO(); if (predictModel == null) { throw new ModelInvokeException("modelEntity is null"); } String modelId = predictModel.getId(); try { List<SampleData> sampleDataList = sampleConstructor.constructSample(TypeA.Predict.name(), modelId, predictTime, itemName, new HashMap<>()); String modelPath = predictModel.getModelpath(); if (modelPath == null) { log.info("模型路径不存在，modelId=" + modelId); return null; } IAILModel newModelBean = composeNewModelBean(predictModel); HashMap<String, Object> settings = getPredictSettingsByModelId(modelId); // 校验setting必须有pyFile，否则可能导致程序崩溃 if (!settings.containsKey(MdkConstant.PY_FILE_KEY)) { 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！"); } if (settings == null) { log.error("模型setting不存在，modelId=" + modelId); return null; } int portLength = sampleDataList.size(); Object[] param2Values = new Object[portLength + 2]; for (int i = 0; i < portLength; i++) { param2Values[i] = sampleDataList.get(i).getMatrix(); } param2Values[portLength] = newModelBean.getDataMap().get("models"); param2Values[portLength + 1] = settings; log.info("####################### 模拟调整 "+ "【itemId:" + predictModel.getItemid() + ",itemName：" + itemName + ",itemNo：" + itemNo + "】 ##########################"); log.info("参数： " + JSON.toJSONString(param2Values)); //IAILMDK.run HashMap<String, Object> modelResult = DllUtils.run(newModelBean, param2Values, predictModel.getMpkprojectid()); //打印结果 log.info("预测模型计算完成：modelId=" + modelId + ",modelName=" + predictModel.getMethodname() + "，modelResult=" + JSON.toJSONString(modelResult)); //判断模型结果 if (!modelResult.containsKey(CommonConstant.MDK_STATUS_CODE) || !modelResult.containsKey(CommonConstant.MDK_RESULT) || !modelResult.get(CommonConstant.MDK_STATUS_CODE).toString().equals(CommonConstant.MDK_STATUS_100)) { throw new ModelResultErrorException("模型结果异常：" + modelResult); } modelResult = (HashMap<String, Object>) modelResult.get(CommonConstant.MDK_RESULT); List<MmItemOutputEntity> itemOutputList = mmItemOutputService.getByItemid(predictModel.getItemid()); Map<MmItemOutputEntity, double[]> predictMatrixs = new HashMap<>(); for (MmItemOutputEntity output : itemOutputList) { if (!modelResult.containsKey(output.getResultstr())) { continue; } OutResultType outResultType = OutResultType.getEumByCode(output.getResultType()); switch (outResultType) { case D1: double[] temp1 = (double[]) modelResult.get(output.getResultstr()); predictMatrixs.put(output, temp1); break; case D2: double[][] temp2 = (double[][]) modelResult.get(output.getResultstr()); double[] tempColumn = new double[temp2.length]; for (int i = 0; i < tempColumn.length; i++) { tempColumn[i] = temp2[i][output.getResultIndex()]; } predictMatrixs.put(output, tempColumn); break; case D: Double temp3 = (Double) modelResult.get(output.getResultstr()); predictMatrixs.put(output, new double[]{temp3}); break; default: break; } } result.setPredictMatrixs(predictMatrixs); result.setModelResult(modelResult); result.setPredictTime(predictTime); } catch (ModelResultErrorException ex) { log.error("模型结果异常", ex); throw ex; } catch (Exception ex) { log.error("调用发生异常，异常信息为:{0}", ex.getMessage()); throw new ModelInvokeException(ex.getMessage()); } return result; } /** * 构造IAILMDK.run()方法的newModelBean参数 * * @param predictModel iailab-module-model/iailab-module-model-biz/src/main/java/com/iailab/module/model/mdk/sample/PredictSampleDataConstructor.java
@@ -30,6 +30,7 @@ import org.springframework.stereotype.Component; import org.springframework.util.CollectionUtils; import java.math.BigDecimal; import java.util.*; import java.util.stream.Collectors; @@ -76,6 +77,8 @@ Map<String, ApiPointDTO> pointMap = sampleInfo.getPointMap(); Map<String, ApiPlanItemDTO> planMap = sampleInfo.getPlanMap(); Map<String, ApiIndItemDTO> indMap = sampleInfo.getIndMap(); int deviationIndex = 0; //对每个爪分别进行计算 for (ColumnItemPort entry : sampleInfo.getColumnInfo()) { //先依据爪内数据项的modelParamOrder进行排序——重写comparator匿名函数 @@ -94,11 +97,31 @@ } } //找出对应的调整值 double[] deviationItem = null; if (sampleInfo.getDeviation() != null && sampleInfo.getDeviation().length > 0) { deviationItem = sampleInfo.getDeviation()[deviationIndex]; } deviationIndex ++; //对每一项依次进行数据查询,然后将查询出的值赋给matrix对应的位置 for (int i = 0; i < entry.getColumnItemList().size(); i++) { try { List<DataValueVO> dataEntityList = getData(entry.getColumnItemList().get(i), pointMap, planMap,indMap); //补全数据 //设置调整值 if (deviationItem != null && deviationItem.length > 0) { logger.info("设置调整值, i = " + i); if (deviationItem[i] <= 0) { continue; } for(int dataKey = 1; dataKey < dataEntityList.size(); dataKey ++) { DataValueVO item = dataEntityList.get(dataKey); item.setDataValue(item.getDataValue() + deviationItem[i]); } } // 补全数据 ColumnItem columnItem = entry.getColumnItemList().get(i); dataEntityList = super.completionData(matrix.length, dataEntityList, columnItem.startTime, columnItem.endTime, columnItem.getParamType(),columnItem.getGranularity());
@@ -44,4 +44,22 @@ } public List<SampleData> constructSample(String typeA, String modelId, Date runTime,String itemName, Map<Integer, Integer> dynamicDataLength, double[][] deviation) throws ModelInvokeException { try { SampleInfoConstructor sampleInfoConstructor = sampleFactory.createSampleInfo(typeA, modelId); SampleInfo sampleInfo = sampleInfoConstructor.prepareSampleInfo(modelId, runTime, dynamicDataLength); sampleInfo.setDeviation(deviation); SampleDataConstructor sampleDataConstructor = sampleFactory.createSampleData(typeA); return sampleDataConstructor.prepareSampleData(sampleInfo); } catch (Exception e) { e.printStackTrace(); log.error("获取模型的算法参数异常",e); throw new ModelInvokeException(MessageFormat.format("{0},Name:{1}", ModelInvokeException.errorGetModelArithParam, itemName)); } } } iailab-module-model/iailab-module-model-biz/src/main/java/com/iailab/module/model/mdk/sample/dto/SampleInfo.java
@@ -33,7 +33,8 @@ private Integer sampleCycle; private BigDecimal[][] deviation; // 调整值 private double[][] deviation; // 所有测点信息，避免重复查询 private Map<String, ApiPointDTO> pointMap; // 所有计划数据信息，避免重复查询 iailab-module-model/iailab-module-model-biz/src/main/resources/application-dev.yaml
@@ -55,7 +55,7 @@ influx-db: org: iailab token: NloIinwybvMwKlJ8SGOAqboXH72EhdQEsnnV7kwtstVu6sbt24LNJ0bVICepeAtl2pxpd1Hj8gDLj9m4hnB7Fw== url: http://127.0.0.1:8086 username: dzd password: qwer1234 token: _338h4Kbu2KQaes5QwAyOz9pTUueXoSF9XmPi8N9oTS1SrhTZVj4J9JfSraUyWA0PfWMZOlf9QWax-USkJQR_A== url: http://172.16.8.200:8086 username: iailab password: iailab2019
