Quality Monitoring Of Injection Molding Process Based On Built-in Sensors

The injection moulding process is a typical batch process with small batches and multiple products.It has obvious intermittent characteristics such as multiple working conditions,multiple time periods,and repeatability.It is also susceptible to interference from various internal and external factors,resulting in unstable quality of injection moulded products.In practical applications,there are still some key problems in the online monitoring of the quality of the injection process:(1)additional information perception equipment will cause increased costs;(2)the modern injection moulding machine itself is equipped with a large number of high-resolution measurement equipment—built-in sensors,the high-quality process data provided by it has not been fully utilized;(3)How to eliminate the adverse effects of the intermittent characteristics of the built-in signal on the quality monitoring model,and improve the model's ability to quickly identify faulty batches.In view of the above problems,this paper proposes two methods for monitoring the quality of injection moulding process based on data driven with built-in signals as monitoring variablesAccording to the key actions of each stage of injection moulding,the paper first selects the appropriate built-in sensors—injection motor torque,plasticizing motor torque,load cell pressure,screw position,load cell and clamping oil pressure.Taking weight as the product quality index,the feasibility and stability of the system are verified by the quality repeat accuracy experimentSecondly,according to the injection moulding process parameter availability standard,by controlling the key processing parameters of each stage of injection moulding,the built-in sensors of each stage of injection moulding are processed and analyzed.It can be seen that the difference between the load cell pressure,the clamping oil pressure load and the no-load,and the torque signal of the plasticized motor that sampling frequency are reduced,can effectively reflect the melt state information of the injection,holding cooling,and plasticizing stages,and can make reasonable and regular changes with the change of the injection moulding conditions In addition,the maximum load element pressure,the plasticized motor torque integral can also better reflect the change of melt quality during the injection moulding processAiming at the problems of low maximum resolution of load cell pressure,low resolution of viscosity index and poor correlation,this paper proposes a multivariable statistical process monitoring method combining built-in signal with MPCA unfolding in batch direction.It analysis the reasons of the multiple working conditions and multiple time periods of the injection moulding process that cause the built-in signals to be unequal in length and unpredictable length,as well as the adverse effects on MPCA.In the case of the same length of train data,the method can still accurately monitor the difference in product quality of ± 0.1942g caused by the increase in mould temperature,which verifies the effectiveness and accuracy of the method under a single working conditionFinally,in order to overcome the cumbersome data preprocessing steps caused by unequal signal length and multiple time periods,reduce the amount of data used for training,reduce the calculation time and storage space,this paper proposes injection process monitoring based on the combination of statistics pattern and k-nearest neighbor The process quality monitoring method(SP-kNN)divides the statistical modulus of the built-in signal into three parts:injection,holding,and plasticization stage,and retains the correlation and process details of each stage of injection moulding.SP-kNN can accurately identify the faulty batch composed of three different working conditions.Compared with FD-kNN's recognition rate of the confirmation set of working condition 1,the SP-kNN recognition rate also reaches 100%,which verifies the multiplexing.Under the circumstances,SP-kNN has better ability to monitor the quality of injection moulding process than FD-kNN.