China
Africa
Explore chapters and articles related to this topic
Soft Sensors: Software-Based Sensors
Published in John G. Webster, Halit Eren, Measurement, Instrumentation, and Sensors Handbook, 2017
The most common application of soft sensors is the prediction of values that cannot be measured online using automated measurements. This may be for technological reasons (e.g., there is no equipment available for the required measurement), economical reasons (e.g., the necessary equipment is too expensive), etc. This often applies to critical values that are related to the final product quality. Soft sensors can in such scenarios provide useful information about the values of interest and in the case when the soft sensor prediction fulfills given standards, it can also be incorporated into the automated control loops of the process [58]. Data-driven soft sensors have been widely used in fermentation, polymerization, and refinery processes.
Force-System Resultants and Equilibrium
Published in Richard C. Dorf, The Engineering Handbook, 2018
Today’s technology allows the development of soft sensors. A soft sensor is a device that does not include any sensing hardware but involves a microprocessor that processes data acquired from a number of devices and combines them by using a mathematical model to produce an estimation of the parameter of interest. Either the traditional models, such as the ARMAX, and NARMAX, or the more exotic neural, fuzzy, and neuro-fuzzy models, are used to obtain the required estimation capability.
Advanced processing of 3D printed biocomposite materials using artificial intelligence
Published in Materials and Manufacturing Processes, 2022
Deepak Verma, Yu Dong, Mohit Sharma, Arun Kumar Chaudhary
This section describes how machine learning is useful for the development of 3D printed biocomposites. As is well known, the application of 3D printing alone has solved most real problems encountered during conventional developments of biocomposites. In addition, some processing parameters still require great attention during a 3D printing process of biocomposites, which may sometimes become quite unfavorable. The introduction of machine learning helped to predict and optimize those parameters at the earlier stages of the prefabrication of biocomposites. A good example of this is given by Mulrennan and coworkers.[66] They utilized the machine learning algorithm to develop a soft sensor that was efficiently useful in predicting the mechanical properties of extruded PLA sheets using an instrumented slit die. This soft sensor is potentially considered as a quality assurance tool, providing real-time feedback to the process/system. This model not only reduces the scrap rate but also lowers the manufacturing costs. The introduction of machine learning-based soft sensors may ultimately be taken as a revolution in smart manufacturing and industry 4.0.
A dynamic soft senor modeling method based on MW-ELWPLS in marine alkaline protease fermentation process
Published in Preparative Biochemistry & Biotechnology, 2021
Xianglin Zhu, Ke Cai, Bo Wang, Khalil Ur Rehman
Soft sensor technology mainly uses a machine-learning algorithm to build mathematical models between the dominant variable and the auxiliary variable, to estimate the value of the dominant variable.[7] Thus it can be seen that the quality of the model is directly related to the accuracy of the final prediction.[8] At the beginning of soft sensor research, most scholars adopt various optimization algorithms to optimize the model parameters, which is a global and off-line soft sensor modeling method.[9] As the fermentation process is time-varying and the working conditions are changing at any time, it is difficult to obtain a high precision with this modeling method, and the model will be invalid over time.[10] In recent years, just-in-time learning (JITL) technique has become the mainstream of soft sensor. It is a local modeling approach which performs well in chemical industry. Minglun Ren et al. adopt locally weighted partial least squares (LWPLS) for industrial soft sensor modeling. Furthermore, particle swarm optimization (PSO) was used to optimize the key parameter (bandwidth h) of LWPLS for increasing its prediction accuracy. The results show that the PSO–LWPLS method has high prediction accuracy over other methods.[11] Yuan et al. proposed an adaptive soft sensor modeling method, which is based on the moving window (MW) and JITL techniques. The results show that the proposed soft sensor framework for nonlinear time-varying processes has high accuracy.[12] Ren and Ma proposed a locally weighted adaptive kernel partial least Square algorithm combined with MWs (LW-AKPLS). In the prediction phase, ensemble learning is utilized to predict the final value after calculating the weights of the sub-model. the algorithm is validated on a set of historical data collected from a coking tower, and the prediction results further demonstrate the effectiveness of the proposed algorithm.[13]
Study on soft sensor modeling method for sign of contaminated fermentation broth in Chlortetracycline fermentation process
Published in Preparative Biochemistry & Biotechnology, 2021
Mei-chun Wang, Xiang Han, Yu-mei Sun, Qiao-yan Sun, Xiang-guang Chen
Abnormal contamination of fermentation broth during Chlortetracycline fermentation can be divided into the contamination of fermentation broth and bacteriophage contamination[5]. After contamination of fermentation broth of Chlortetracycline fermentation process, a large number of bacteria will metabolize and multiply in large quantities, thus consuming a large number of carbon sources, resulting in rapid reduction of total sugar content[6]. After the bacteriophage contaminated the fermentation process of Chlortetracycline, the bacteriophage parasitizes and kills the normal target bacteria, causing a large number of death of the normal target bacteria, reducing the consumption rate of total sugar content, resulting in a high total sugar content. Experts and scholars in China have studied that data-driven soft sensor model can predict online variable data and estimate its state. For example, Cheng and other scholars have studied the soft sensor modeling method based on data-driven model for processes with non-linear characteristics[7]. Yang has studied soft sensor algorithm based on artificial neural network[8]. Chen further studied the soft sensing method of multi-model mixing and applied it to penicillin fermentation process[9], Xu realized the soft sensor of fermentation process by hybrid modeling of mechanism model and data-driven model[10]. Yang has studied the application experiment of parallel mixing model in fermentation process[11]. These soft sensor methods have their own characteristics and can be used for soft sensor of related variables in different industrial processes. Yu introduced ten of the major modeling methods of soft sensing, they include mechanism modeling, regression analysis, state estimation, pattern recognition, artificial neural network, fuzzy mathematics, and SVM- or kernel-function-based method, process tomography, correlative analysis and non-linear system information processing technology[12].