模型的保存，加载及清理

	修改：     RadomForestSAC/FeatureExtractor.cpp
	修改：     RadomForestSAC/RamdomForestCalculate.cpp
	修改：     RadomForestSAC/RandomForestClassifier.cpp
	修改：     package/E3/SigerCalculation

SigerCalculation/RadomForestSAC/FeatureExtractor.cpp

@@ -5,7 +5,6 @@
 #include <cmath>
 #include <iostream>
 #include <complex>
-#include <sys/time.h>
 #include "spdlog/spdlog.h"
 #include "common.h"
 
@@ -51,9 +50,6 @@ SignalFeatures FeatureExtractor::extract_features(const std::vector<float>& sign
         return features;
     }
 
-    unsigned long elapsed_time = 0;
-    struct timeval start,end;
-    gettimeofday(&start, NULL);
     std::vector<float> processed_signal = signal;
 
     // 1. 高通滤波处理 - 使用预定义系数
@@ -88,9 +84,6 @@ SignalFeatures FeatureExtractor::extract_features(const std::vector<float>& sign
         }
     }
 
-    gettimeofday(&end, NULL);
-    elapsed_time = (end.tv_sec - start.tv_sec) * 1000L + (end.tv_usec - start.tv_usec)/1000;
-    SPDLOG_LOGGER_DEBUG(spdlog::get("logger"),"[extract_features] elapsed_time: {}", elapsed_time);
     return features;
 }
 

SigerCalculation/RadomForestSAC/RamdomForestCalculate.cpp

@@ -36,11 +36,11 @@ void delete_excess_rf_models()
                     ++done;
             }
         }
+
+        SPDLOG_LOGGER_DEBUG(spdlog::get("logger"),"[delete_rf_models] to_delete={} done={}", to_delete, done);
     }
 
     closedir(dir);
-    if (to_delete > 0)
-        SPDLOG_LOGGER_DEBUG(spdlog::get("logger"),"[delete_rf_models] to_delete={} done={}", to_delete, done);
 }
 
 int pickPriorityValue(double xpred, double ypred, double zpred)
@@ -245,6 +245,9 @@ int RamdomForestCalculate::Save_SACDATA(const Dc_SacData &sacdata)
     QString FilePath = QString("/home/pi/SigerTMS/stream/SAC/%1").arg(sacdata.sacFile);
 
     //step1: 读取二进制原始文件
+    unsigned long elapsed_time = 0;
+    struct timeval start,end;
+    gettimeofday(&start, NULL);
     VibrationDataReader reader;
     if (reader.readFileRaw(FilePath)) {
         //reader.printDataInfo();
@@ -255,6 +258,9 @@ int RamdomForestCalculate::Save_SACDATA(const Dc_SacData &sacdata)
     raw = reader.convertToRawFormat();
     for (int i=0; i<3; ++i)
         fea[i] = extractor.extract_features(raw[i]);
+    gettimeofday(&end, NULL);
+    elapsed_time = (end.tv_sec - start.tv_sec) * 1000L + (end.tv_usec - start.tv_usec)/1000;
+    SPDLOG_LOGGER_DEBUG(spdlog::get("logger"),"[extract_features] elapsed_time: {}", elapsed_time);
 
     //step2: 预测报警
     int valid_num = 0;
@@ -355,7 +361,8 @@ bool RamdomForestCalculate::queryRecordsByLabel(const QString &dbFileName,
 
         QSqlQuery query(db);
         if (!query.exec(sql)) {
-            qDebug() << "Query failed:" << query.lastError().text();
+            QString error = query.lastError().text();
+            SPDLOG_LOGGER_ERROR(spdlog::get("logger"), "Query failed: {}", error.toUtf8().constData());
             db.close();
             QSqlDatabase::removeDatabase(conn);
             return false;
@@ -372,7 +379,7 @@ bool RamdomForestCalculate::queryRecordsByLabel(const QString &dbFileName,
             record.y = query.value("y").toString();
             record.z = query.value("z").toString();
 
-            if (record.target.isEmpty() || record.startTime.isEmpty() || record.startTime.isEmpty() ||
+            if (record.target.isEmpty() || record.startTime.isEmpty() || record.endTime.isEmpty() ||
                     record.x.isEmpty() || record.y.isEmpty() || record.z.isEmpty())
                 continue;
 
@@ -388,6 +395,8 @@ bool RamdomForestCalculate::queryRecordsByLabel(const QString &dbFileName,
                     int replace_index = std::rand() % max_normal_per_target;
                     target_pair.second[replace_index] = record;
                 }
+
+                //SPDLOG_LOGGER_ERROR(spdlog::get("logger"), "target={} size={}", record.target.toStdString().c_str(), target_pair.second.size());
             }
         }
 

SigerCalculation/RadomForestSAC/RandomForestClassifier.cpp

@@ -6,7 +6,7 @@
 
 RandomForestClassifier::RandomForestClassifier(std::string target, int initial_dim, int label_num)
 {
-    target = m_target;
+    m_target = target;
     feature_dim = initial_dim;
     label_dim = label_num;
 }
@@ -202,10 +202,10 @@ bool RandomForestClassifier::train(const std::vector<std::vector<float>>& X, con
     {
         std::string filename = (!m_target.empty()) ? (m_target + "_rf_model.dat") : "rf_model.dat";
         serialize(filename) << scaler_mean << scaler_std << forest;
-        DEBUG_PRINT("save random-forest model success\n");
+        SPDLOG_LOGGER_DEBUG(spdlog::get("logger"), "save {} success", filename.c_str());
     }
     catch(std::exception &e) {
-        DEBUG_PRINT("save random-forest model failed: %s\n", e.what());
+        SPDLOG_LOGGER_DEBUG(spdlog::get("logger"), "load model failure: {}", e.what());
         return false;
     }
 
@@ -279,7 +279,7 @@ void RandomForestClassifier::evaluate(const std::vector<sample_type>& X_test, co
     for (int true_val = 0; true_val < label_dim; ++true_val) {
         if (ret < maxlen)
             ret += snprintf(&ptr[ret], maxlen-ret, "真实 %d: ", true_val);
-        for (int pred_val = 0; pred_val < label_dim; ++pred_val) {
+        for (int pred_val = 0; pred_val < label_dim; ++pred_val){
             if (ret < maxlen)
                 ret += snprintf(&ptr[ret], maxlen-ret, "%3d ", conf_matrix[true_val][pred_val]);
         }
@@ -312,9 +312,10 @@ double RandomForestClassifier::predict(const std::vector<float>& sample)
             std::string filename = (!m_target.empty()) ? (m_target + "_rf_model.dat") : "rf_model.dat";
             deserialize(filename) >> scaler_mean >> scaler_std >> forest;
             feature_dim = scaler_mean.size();
+            SPDLOG_LOGGER_DEBUG(spdlog::get("logger"), "load {} success", filename.c_str());
         }
         catch (const std::exception& e) {
-            SPDLOG_LOGGER_DEBUG(spdlog::get("logger"), "load model failure: {}\n", e.what());
+            SPDLOG_LOGGER_DEBUG(spdlog::get("logger"), "load model failure: {}", e.what());
             return -1;
         }
     }