Research On Indirect Measurement Method And System Design Implementation Of Medium Frequency Furnace Charge Temperature

With the continuous development of induction heating and semiconductor technology,medium-frequency electric furnaces have been widely applied in the smelting industry due to their advantages of high efficiency,energy saving,and stable operation.Temperature is the most fundamental parameter in the smelting process.It can help furnace operators understand the state of the metal and carry out corresponding smelting operations based on the smelting process,thereby ensuring the quality of the metal,improving production efficiency,and reducing production costs.However,in most smelting plants in our country,temperature measurement still relies on the use of ordinary thermocouples,which have the disadvantages of short lifespan,high cost,and inability to continuously measure temperature.In this paper,a new method for indirectly measuring the temperature of the medium-frequency furnace charge is proposed,based on the characteristic of the furnace charge resistance increasing with temperature.By collecting the furnace charge reflection resistance and furnace charge temperature related to resistance,and relying on the Transformer-LSTM hybrid model,this method provides a new approach for indirect temperature measurement of the medium-frequency furnace charge.Firstly,this paper analyzes the relationship between electrical parameters of the load circuit in a medium-frequency furnace and utilizes the relationship between metal resistivity and temperature to analyze the impact of temperature on the resistance of the metal charge.Field measurements of the load circuit parameters in the medium-frequency furnace are conducted,and a simulated medium-frequency furnace load circuit is built using Multisim simulation software to further validate the significant impact of temperature on the resistance of the metal charge in the load circuit of the medium-frequency furnace.A new method for indirectly measuring the temperature of the medium-frequency furnace charge using reflection resistance is proposed.Secondly,based on the proposed temperature measurement method,a data acquisition system is designed for the parallel resonant medium-frequency furnace.The system collects the load circuit voltage,current,and furnace charge temperature.Using devices such as platinum-rhodium thermocouples,current transformers,and AD210 BN,a data conversion circuit is designed to convert the load circuit voltage,current,and furnace charge temperature into voltage signals that can be acquired by the AD acquisition circuit.The data signals are collected through the configuration of the processor circuit and the AD acquisition chip.Suitable devices are chosen to design the power circuit according to the power supply requirements of each module.The schematic and PCB layout of each circuit in the data acquisition system are drawn,verifying the effectiveness of the data acquisition system and achieving the collection of resonance circuit voltage,current,and furnace charge temperature data during the medium-frequency furnace smelting process.Thirdly,the collected voltage and current data from the data acquisition system are subjected to discrete Fourier transform to obtain the fundamental components of the voltage and current data.By deriving the reflection resistance and processing the collected temperature data,a dataset of temperature and reflection resistance is obtained,providing data support for subsequent algorithm validation.A temperature prediction method based on the Transformer-LSTM hybrid model is proposed,and the model is constructed and parameterized.The construction of the dataset in this paper is described.In order to verify the advantages of the proposed hybrid model,LSTM and Transformer models are also constructed and parameterized.The models are trained using the dataset,and the training results of the Transformer-LSTM hybrid model,LSTM model,and Transformer model are compared one by one.Through data comparison,the advantages of the proposed hybrid model in temperature prediction using reflection resistance are further demonstrated.The accuracy of temperature prediction can reach 98.86%,thus proving the effectiveness and accuracy of the algorithm.Finally,this paper utilizes the data acquisition system to collect voltage and current data.Using Py Qt5,an upper computer is designed to configure the serial port and receive data sent from the data acquisition system.The received data is further processed to obtain the reflection resistance value,which serves as input for the proposed Transformer-LSTM hybrid model.By processing the reflection resistance data,the furnace charge temperature data can be obtained,realizing the design and implementation of the medium-frequency furnace charge temperature measurement system.Finally,a comparison is made between the temperature data obtained from the medium-frequency furnace temperature measurement system and the temperature data obtained from the platinum-rhodium thermocouple.The built medium-frequency furnace temperature measurement system achieves a temperature accuracy of 96.37%.This further demonstrates the effectiveness of the indirect temperature measurement method for the medium-frequency furnace charge temperature based on reflection resistance,which can provide relatively accurate temperature reference data for subsequent smelting operations in the melting process.