package com.iailab.module.model.mcs.pre.service.impl;
|
|
import com.baomidou.mybatisplus.core.metadata.IPage;
|
import com.iailab.framework.common.pojo.PageResult;
|
import com.iailab.framework.common.service.impl.BaseServiceImpl;
|
import com.iailab.module.data.api.point.DataPointApi;
|
import com.iailab.module.data.api.point.dto.ApiPointDTO;
|
import com.iailab.module.data.api.point.dto.ApiPointValueDTO;
|
import com.iailab.module.data.api.point.dto.ApiPointValueQueryDTO;
|
import com.iailab.module.model.common.enums.DataTypeEnum;
|
import com.iailab.module.model.influxdb.pojo.InfluxModelResultLastSimPOJO;
|
import com.iailab.module.model.influxdb.pojo.InfluxModelResultPOJO;
|
import com.iailab.module.model.influxdb.service.InfluxDBService;
|
import com.iailab.module.model.influxdb.vo.InfluxModelResultVO;
|
import com.iailab.module.model.mcs.pre.dao.MmPredictAutoAdjustConfigDao;
|
import com.iailab.module.model.mcs.pre.entity.MmPredictAutoAdjustConfigEntity;
|
import com.iailab.module.model.mcs.pre.enums.AutoAdjustTriggerRuleEnum;
|
import com.iailab.module.model.mcs.pre.enums.AutoAdjustValueRuleEnum;
|
import com.iailab.module.model.mcs.pre.service.MmPredictAutoAdjustConfigService;
|
import com.iailab.module.model.mcs.pre.service.MmPredictItemService;
|
import com.iailab.module.model.mcs.pre.vo.MmPredictAutoAdjustConfigPageReqVO;
|
import com.iailab.module.model.mcs.pre.vo.MmPredictAutoAdjustConfigVO;
|
import com.iailab.module.model.mdk.vo.ItemVO;
|
import lombok.extern.slf4j.Slf4j;
|
import org.springframework.beans.factory.annotation.Autowired;
|
import org.springframework.stereotype.Service;
|
import org.springframework.util.CollectionUtils;
|
|
import java.math.BigDecimal;
|
import java.util.*;
|
import java.util.function.Predicate;
|
import java.util.stream.Collectors;
|
|
/**
|
* @author PanZhibao
|
* @Description
|
* @createTime 2024年11月19日
|
*/
|
@Slf4j
|
@Service
|
public class MmPredictAutoAdjustConfigServiceImpl extends BaseServiceImpl<MmPredictAutoAdjustConfigDao, MmPredictAutoAdjustConfigEntity>
|
implements MmPredictAutoAdjustConfigService {
|
|
@Autowired
|
private MmPredictItemService mmPredictItemService;
|
@Autowired
|
private InfluxDBService influxDBService;
|
@Autowired
|
private DataPointApi dataPointApi;
|
|
@Override
|
public PageResult<MmPredictAutoAdjustConfigVO> page(MmPredictAutoAdjustConfigPageReqVO params) {
|
IPage<MmPredictAutoAdjustConfigVO> page = baseDao.selectPage(params);
|
return new PageResult<>(page.getRecords(), page.getTotal());
|
}
|
|
@Override
|
public MmPredictAutoAdjustConfigEntity getInfo(String id) {
|
return baseDao.selectById(id);
|
}
|
|
@Override
|
public List<MmPredictAutoAdjustConfigEntity> getByCode(String code) {
|
return baseDao.selectList(MmPredictAutoAdjustConfigEntity::getConfigCode,code,MmPredictAutoAdjustConfigEntity::getIsEnable,1);
|
}
|
|
@Override
|
public void create(MmPredictAutoAdjustConfigEntity entity) {
|
entity.setId(UUID.randomUUID().toString());
|
entity.setCreateTime(new Date());
|
baseDao.insert(entity);
|
}
|
|
@Override
|
public void update(MmPredictAutoAdjustConfigEntity entity) {
|
baseDao.updateById(entity);
|
}
|
|
@Override
|
public void delete(String id) {
|
baseDao.deleteById(id);
|
}
|
|
@Override
|
public boolean autoAdjustByCode(String configCode,long adjustStartTime) {
|
log.info("开始自动调整：configCode：" + configCode + ",adjustStartTime:" + new Date(adjustStartTime));
|
// 查询调整配置
|
List<MmPredictAutoAdjustConfigEntity> configEntityList = getByCode(configCode);
|
if (CollectionUtils.isEmpty(configEntityList)) {
|
log.info("自动调整失败原因：configEntityList为空");
|
return false;
|
}
|
|
// 根据outputId分组
|
Map<String, List<MmPredictAutoAdjustConfigEntity>> outputIdMap = configEntityList.stream().collect(Collectors.groupingBy(MmPredictAutoAdjustConfigEntity::getOutputId));
|
for (Map.Entry<String, List<MmPredictAutoAdjustConfigEntity>> entry : outputIdMap.entrySet()) {
|
String outputId = entry.getKey();
|
// 查询调整用户adjustStartTime 至 adjustStartTime - 预测长度 * 预测粒度 范围的值
|
Calendar calendar = Calendar.getInstance();
|
calendar.setTimeInMillis(adjustStartTime);
|
Date endTime = calendar.getTime();
|
ItemVO item = mmPredictItemService.getItemByOutPutId(outputId);
|
if (item == null) {
|
log.info("自动调整失败原因：getItemByOutPutId为null，outputId：" + outputId);
|
continue;
|
}
|
calendar.add(Calendar.SECOND,item.getPredictLength() * item.getGranularity() * -1);
|
Date startTime = calendar.getTime();
|
|
// 获取预测历史结果
|
InfluxModelResultPOJO pojo = new InfluxModelResultPOJO();
|
pojo.setType(DataTypeEnum.FLOAT_LAST_BAK.getCode());
|
pojo.setOutPutId(outputId);
|
List<InfluxModelResultVO> influxModelResult = influxDBService.queryModelResults(pojo, new Date(adjustStartTime), new Date(adjustStartTime));
|
if (CollectionUtils.isEmpty(influxModelResult)) {
|
log.info("自动调整失败原因：预测历史结果为空。itemNo：" + item.getItemNo() + "，itemName:" + item.getItemName() + "，outputId:" + outputId + "，time:" + adjustStartTime);
|
continue;
|
}
|
|
// 计算所有影响用户的最终调整值
|
Double finalAdjustValue = 0.0;
|
for (MmPredictAutoAdjustConfigEntity configEntity : entry.getValue()) {
|
Double adjustValue = null;
|
// 查询影响用户历史值
|
ApiPointValueQueryDTO queryDTO = new ApiPointValueQueryDTO();
|
ApiPointDTO pointInfo = dataPointApi.getInfoById(configEntity.getPointId());
|
queryDTO.setPointNo(pointInfo.getPointNo());
|
queryDTO.setEnd(endTime);
|
queryDTO.setStart(startTime);
|
List<ApiPointValueDTO> apiPointValueDTOS = dataPointApi.queryPointHistoryValue(queryDTO);
|
if (CollectionUtils.isEmpty(apiPointValueDTOS)) {
|
log.info("影响用户[" + pointInfo.getPointName() + "]调整失败原因：测点数据长度为0。queryDTO：" + queryDTO);
|
continue;
|
}
|
// 过滤掉-2
|
apiPointValueDTOS = apiPointValueDTOS.stream().filter(e -> !Double.valueOf(e.getV()).equals(-2.0)).collect(Collectors.toList());
|
if (CollectionUtils.isEmpty(apiPointValueDTOS)) {
|
log.info("影响用户调整失败原因：过滤掉-2之后测点数据长度为0。queryDTO：" + queryDTO);
|
continue;
|
}
|
|
// 触发规则
|
AutoAdjustTriggerRuleEnum triggerRuleEnum = AutoAdjustTriggerRuleEnum.fromCode(configEntity.getTriggerRule());
|
// 判断是否符合触发条件 并计算调整值
|
switch (triggerRuleEnum) {
|
case SLOPE:
|
// 计算每个△t的斜率，任意一个大于触发值则认为该区间有调整
|
Calendar slopeCalendar = Calendar.getInstance();
|
slopeCalendar.setTime(startTime);
|
Date slopeStartTime = slopeCalendar.getTime();
|
slopeCalendar.add(Calendar.MINUTE,configEntity.getT());
|
Date slopeEndTime = slopeCalendar.getTime();
|
if (slopeEndTime.after(endTime)) {
|
log.info("影响用户[" + pointInfo.getPointName() + "]调整失败原因：△t设置过大，大于模型预测长度 * 预测粒度。△t：" + configEntity.getT());
|
continue;
|
}
|
while (!slopeEndTime.after(endTime)) {
|
//计算斜率
|
//△t开始时间测点值
|
Date finalSlopeStartTime = slopeStartTime;
|
Optional<ApiPointValueDTO> startOptional = apiPointValueDTOS.stream().filter(apiPointValueDTO -> apiPointValueDTO.getT().equals(finalSlopeStartTime)).findFirst();
|
//△t结束时间测点值
|
Date finalSlopeEndTime = slopeEndTime;
|
Optional<ApiPointValueDTO> endOptional = apiPointValueDTOS.stream().filter(e -> e.getT().equals(finalSlopeEndTime)).findFirst();
|
if (startOptional.isPresent() && endOptional.isPresent()) {
|
ApiPointValueDTO startPointValue = startOptional.get();
|
ApiPointValueDTO endPointValue = endOptional.get();
|
// 计算斜率
|
double slope = BigDecimal.valueOf(endPointValue.getV() - startPointValue.getV()).divide(BigDecimal.valueOf(configEntity.getT())).doubleValue();
|
// 斜率大于等于触发值则进行调整
|
if (Double.valueOf(Math.abs(slope)).compareTo(configEntity.getTriggerValue()) >= 0) {
|
// 计算调整值 并跳出循环
|
adjustValue = AutoAdjustValueRuleEnum.getAdjustValue(configEntity.getAdjustValueRule(), apiPointValueDTOS);
|
log.info("计算调整值：" + adjustValue + "，斜率：" + slope + "，pointNo：" + pointInfo.getPointNo() + "，pointName：" + pointInfo.getPointName() + "，slopeStartTime：" + slopeStartTime + "，slopeEndTime：" + slopeEndTime);
|
break;
|
}
|
log.info("斜率不满足条件，斜率：" + slope);
|
}
|
// 下一个△t
|
slopeStartTime = slopeCalendar.getTime();
|
slopeCalendar.add(Calendar.MINUTE,configEntity.getT());
|
slopeEndTime = slopeCalendar.getTime();
|
}
|
break;
|
case AVERAGE_GAP:
|
// 计算每两个△t的平均差，任意一个大于触发值则认为该区间有调整
|
Calendar averageCalendar = Calendar.getInstance();
|
averageCalendar.setTime(startTime);
|
Date averageStartTime = averageCalendar.getTime();
|
averageCalendar.add(Calendar.MINUTE,configEntity.getT());
|
Date averageMiddleTime = averageCalendar.getTime();
|
averageCalendar.add(Calendar.MINUTE,configEntity.getT());
|
Date averageEndTime = averageCalendar.getTime();
|
|
if (averageEndTime.after(endTime)) {
|
log.info("影响用户[" + pointInfo.getPointName() + "]调整失败原因：△t设置过大，△t*2大于模型预测长度 * 预测粒度。△t：" + configEntity.getT());
|
continue;
|
}
|
while (!averageEndTime.after(endTime)) {
|
//计算均值差
|
//前△t测点平均值
|
Date finalAverageStartTime = averageStartTime;
|
Date finalAverageMiddleTime = averageMiddleTime;
|
OptionalDouble startAverage = apiPointValueDTOS.stream().filter(e -> e.getT().after(finalAverageStartTime) && !e.getT().after(finalAverageMiddleTime)).mapToDouble(ApiPointValueDTO::getV).average();
|
//后△t测点平均值
|
Date finalAverageEndTime = averageEndTime;
|
OptionalDouble endAverage = apiPointValueDTOS.stream().filter(e -> e.getT().after(finalAverageMiddleTime) && !e.getT().after(finalAverageEndTime)).mapToDouble(ApiPointValueDTO::getV).average();
|
if (startAverage.isPresent() && endAverage.isPresent()) {
|
double averageGapValue = startAverage.getAsDouble() - endAverage.getAsDouble();
|
// 均值差，大于等于触发值则进行调整
|
if (Double.valueOf(Math.abs(averageGapValue)).compareTo(configEntity.getTriggerValue()) >= 0) {
|
// 计算调整值 并跳出循环
|
adjustValue = AutoAdjustValueRuleEnum.getAdjustValue(configEntity.getAdjustValueRule(), apiPointValueDTOS);
|
log.info("计算调整值：" + adjustValue + "，均值差：" + averageGapValue + "，pointNo：" + pointInfo.getPointNo() + "，pointName：" + pointInfo.getPointName() + "，averageStartTime：" + averageStartTime + "，averageMiddleTime：" + averageMiddleTime + "，averageEndTime：" + averageEndTime);
|
break;
|
}
|
log.info("均值差不满足条件，均值差：" + averageGapValue);
|
}
|
|
// 下一个△t
|
averageStartTime = averageMiddleTime;
|
averageMiddleTime = averageEndTime;
|
averageCalendar.add(Calendar.MINUTE,configEntity.getT());
|
averageEndTime = averageCalendar.getTime();
|
}
|
break;
|
default:
|
log.info("影响用户[" + pointInfo.getPointName() + "]调整失败原因：未知触发规则,triggerRule" + configEntity.getTriggerRule());
|
continue;
|
}
|
if (adjustValue == null) {
|
log.info("影响用户[" + pointInfo.getPointName() + "]调整失败原因：未达到触发条件");
|
continue;
|
}
|
|
// 调整系数
|
adjustValue = adjustValue * configEntity.getAdjustCoefficient();
|
// 调整方向
|
adjustValue = adjustValue * configEntity.getAdjustDirection();
|
// 累加到最终调整值
|
finalAdjustValue += adjustValue;
|
}
|
// 执行调整
|
if (finalAdjustValue.equals(0.0)) {
|
log.info("自动调整失败原因：finalAdjustValue为0，outputId：" + outputId + "，configCode:" + configCode);
|
continue;
|
}
|
List<InfluxModelResultPOJO> lastList = new ArrayList<>();
|
for (InfluxModelResultVO resultVO : influxModelResult) {
|
InfluxModelResultLastSimPOJO adjustPojo = new InfluxModelResultLastSimPOJO();
|
// 设置新的调整值
|
adjustPojo.setValue(Double.parseDouble(resultVO.getValue().toString()) + finalAdjustValue);
|
adjustPojo.setTimestamp(resultVO.getTimestamp());
|
adjustPojo.setOutPutId(outputId);
|
lastList.add(adjustPojo);
|
}
|
// 相同时间直接覆盖旧值
|
influxDBService.asyncWriteModelResults(lastList);
|
log.info("t+l自动调整。configCode:" + configCode + ",adjustValue:" + finalAdjustValue + "，itemNo：" + item.getItemNo() + "，itemName：" + item.getItemName() + "，outputId：" + outputId + ",adjustTime：" + adjustStartTime);
|
}
|
|
return true;
|
}
|
}
|
