选矿过程关键设备故障诊断和预测数据集

针对位于项目示范单位金川集团选矿厂一选车间碎磨系统关键设备及流程收集了用于关键设备故障诊断和预测技术开发的原始数据，包括用于带式输运机故障诊断、矿用皮带秤故障诊断、基于渣浆泵、旋流器的磨矿分级系统故障诊断以及球磨机、带式输运机故障预测技术开发的原始数据，以及为了完成特定技术目标而进行的对于原始数据的分析数据。基于上述原始数据和分析数据形成了“选矿过程关键设备故障诊断和预测数据集”。其中，（1）带式输运机故障诊断原始数据由1、2号带式输运机驱动电流测量值和带式输运机启停状态组成（数据量：105MB），基于该原始数据使用子空间辨识方法生成分析数据（数据量：148MB），分析数据用于诊断带式输运机故障；（2）矿用皮带秤故障诊断原始数据主要由矿用皮带秤测量值和对应的带式输运机驱动电流测量值组成（数据量:95.9MB），来源于矿用皮带秤和对应的带式输运机电流传感器，基于该原始数据使用支持向量机方法生成分析数据(数据量:188MB)，分析数据用于诊断矿用皮带秤故障；（3）基于渣浆泵、旋流器的磨矿分级系统故障诊断原始数据主要由渣浆泵的频率、旋流器的给矿浓度、旋流器的给矿流量、旋流器的给矿压力、泵池的液位等数据组成（数据量：182MB），来源于磨矿分级系统的浓度、流量、压力、液位等传感器，基于该原始数据使用残差卷积神经网络自编码器模型生成分析数据（数据量：220MB），分析数据用于诊断磨矿分级系统故障；（4）球磨机和带式输运机的故障预测原始数据主要由振动、温度数据组成（数据量：523MB），来源于振动、温度传感器，基于该原始数据使用频谱分析、包络谱分析、小波降噪、倒谱分析等方法产生了分析数据（数据量：1.9GB），分析数据用于球磨机和带式输运机的故障预测。

Raw data was collected for the development of key equipment fault diagnosis and prognosis technologies from the critical equipment and processes of the crushing and grinding system in the No.1 Workshop of Jinchuan Group Concentrator Plant, which serves as the project demonstration unit. This raw data covers fault diagnosis for belt conveyors, fault diagnosis for mining belt weighers, fault diagnosis for the grinding classification system based on slurry pumps and hydrocyclones, and fault prognosis for ball mills and belt conveyors, as well as analytical data generated from the raw data to achieve specific technical objectives. Based on the aforementioned raw and analytical data, the "Key Equipment Fault Diagnosis and Prognosis Dataset for Mineral Processing Processes" is formed.
1. The raw data for belt conveyor fault diagnosis consists of driving current measurements of No.1 and No.2 belt conveyors and the start/stop status of the belt conveyors (data volume: 105 MB). Analytical data (data volume: 148 MB) is generated from this raw data using the subspace identification method, which is used for belt conveyor fault diagnosis.
2. The raw data for mining belt weigher fault diagnosis mainly consists of mining belt weigher measurements and corresponding driving current measurements of the belt conveyor (data volume: 95.9 MB), sourced from current sensors of the mining belt weigher and the associated belt conveyor. Analytical data (data volume: 188 MB) is generated from this raw data using the Support Vector Machine (SVM) method, which is used for mining belt weigher fault diagnosis.
3. The raw data for fault diagnosis of the grinding classification system based on slurry pumps and hydrocyclones mainly includes parameters such as slurry pump frequency, hydrocyclone feed concentration, hydrocyclone feed flow rate, hydrocyclone feed pressure, and sump level (data volume: 182 MB), sourced from concentration, flow rate, pressure, and level sensors of the grinding classification system. Analytical data (data volume: 220 MB) is generated from this raw data using the Residual Convolutional Neural Network Autoencoder model, which is used for fault diagnosis of the grinding classification system.
4. The raw data for fault prognosis of ball mills and belt conveyors mainly consists of vibration and temperature data (data volume: 523 MB), sourced from vibration and temperature sensors. Analytical data (data volume: 1.9 GB) is generated from this raw data using methods including spectral analysis, envelope spectrum analysis, wavelet denoising, and cepstrum analysis, which is used for fault prognosis of ball mills and belt conveyors.

搜集汇总

背景与挑战

背景概述

该数据集包含选矿过程中关键设备（如带式输运机、矿用皮带秤、渣浆泵等）的故障诊断和预测数据，数据量1.36GB，来源于金川集团选矿厂，用于支持故障诊断和预测技术开发。