In modern industry, enterprises are faced with great pressure of competition. Whether they win or not depends mainly on the efficiency of production and the quality of the products. However, with the continuous development of science and technology, the production line gradually tends to be more mechanized and intelligent, but the artificial participation degree is still relatively high in the assembly line. Therefore, the man-machine system of the assembly line affects the production efficiency and the quality of the product directly, and it also affects the economic benefit of enterprises.

There are many research works that study the reliability of production line equipment. Yang[1] set up a rate model for equipment failure on the serial production line. The relationship between the reliability of the production line and the maintenance time of the equipment was deduced. The reliability of the production line is related not only to equipment failure, but also to the maintenance level, the actual cache amount in each buffer zone, the productivity of the bottle neck unit, and the location of the equipment in the production line.

Li [2]focused on the reliability allocation of engine production line considering caching factors, established the reliability block diagram of reference production line and the fuzzy cost hierarchy analysis model, and used the interval number theory to determine the weight of production line cost factors. At the same time, the cost function theory was introduced to establish the production line reliability allocation model. To ensure the reliability, Liang and others established a production unit optimization model based on the preventive maintenance policy and improved the optimization model based on the reliability constraints [3].

Cui [4] established a production line fault database through the domestic automobile engine cylinder head production line tracking fault conditions and conducted a component-level failure mode and impact analysis. Fanpresented the idea that cognitive heterogeneity causes operation time variability and high human error rates at human-based stations of assembly lines. This may affect the performance of production activities, the design process of which depends on the reliable estimate of operation time. To quantify operator cognition, the author measured the human factors’ complexity of human-based stations using an information entropy approach inaspect of operation time. Initially, the influence of the operators’ cognition on the operation time wasanalysed. Based on the analysis, the operation time model that considers cognition was presented. The operation time model was then modified by applying a correction method using the behaviour correction factor, and, finally, the human factors’ complexity measurement model was built. A case study was also used to illustrate the application of the proposed method in order to measure the influence of human cognition on the performance of the human-based station [5].

Zhang [6] adopted the reliability recovery factor to describe the reliability of evolution before and after the equipment maintenance and built a reliability based on the optimization target of minimizing the total cost of the series-parallel production system to maintain the model. Through the concrete analysis (PFMEA and fault tree strategy FTA) that affects the reliability of key equipment, Gu [7] developed the best preventive maintenance cycle for key equipment by using the potential failure theory.

Meng[8] simplified the equipment of the production line into the form of the production unit and calculated the failure rate and the repair rate of each unit. This work can be used to improve the reliability project of the PC production line and provide reference for enterprises to improve the availability of production lines. Through the finite element method, the key parts such as feed screw and bolt were analysed. In view of the design shortage of the feeding screw, the corresponding improvement suggestions were put forward to improve the reliability of the mixer [9].

Human error is more complex than equipment. Different from other production lines on the equipment or the reliability analysis of the whole production line, the reliability of the assembly line is gradually valued by people [10]. Wu [11]combined the two models of HCR and THERP, using the HCR model to calculate the probability of human error in the diagnosis process and the THERP model to calculate the probability of human error in the operation process. A more comprehensive and accurate description of errors in employee behaviour has been quantified.

Zhang [12] proposed the concept of observational reliability and established a series of human factors reliability evaluation indexes that is convenient for statistics and assignment. Zhang established a human factor reliability evaluation model based on RBF neural network for coal miners. Bai[13] studied the human factors on the TWT production line and used human engineering knowledge to improve the environment and optimize the assembly room.

Starting from the behaviour formation factors of assembly operation, Wu[14] quantitatively identified the main factors of operator behaviour formation through the set-value statistics and comprehensive evaluation function of behaviour formation factors and determined the main factors that affect the operator reliability.

The above works show that the reliability of the assembly line is mainly studied from the reliability of equipment and people. The reliability of the equipment is mainly reflected in whether it can run stably without hindering the smooth operation of the production line and whether it can be repaired in time when the equipment fails so that the equipment can be put back into production. To a certain extent, the reliability of the equipment can be measured objectively and maintained regularly through measured data and calculated results to avoid affecting production.

Infact, most production accidents that affect the efficiency and stability of the production line are due to human errors. However, due to the uncertainty and instability of humans, the possibility of errors in assembly or production of production lines is highly volatile. Especially now, most production lines are mixed flow production lines, and it is common that parts are wrongly taken or omitted in production.

The direct cause of the insecurity of the material causes may also be indirectly caused by human factors. Therefore, improving people’s reliability is directly related to the production line efficiency and product quality. In this paper, the improved LEC method is combined with the CREAM method and is used to study the reliability of human factors that affect the production efficiency and stability of the assembly line.

The Cognitive Reliability and Error Analysis Method (CREAM) is one of the second generations of human reliability analysis methods and involves the analysis of the task environment to directly determine the probability of human error. The main function of CREAM prediction analysis is to predict the probability of failure of a person’s cognitive activity task.

The CREAM method uses the Contextual Control Model (COCOM) as the basis of a cognitive model. CREAM considers that the basic probabilities of human errors can be roughly determined under four different control modes [15]: scrambled mode, opportunistic mode, tactical mode, and strategic mode, as shown in Table 1.

In terms of reliability assessment, the Basic Law and Extension Method are two methods of CREAM [16]. For the Basic Law, determining the mode of control is the key link. The basic idea of extended method predictive analysis is to analyze the cognitive activities and possible cognitive failures of people in the process of completing tasks. On the basis ofobtaining the basic value of cognitive function failure probability, the CPC factor level of the situation environment in the study was obtained. The basic value is modified to predict the probability of failure when a person completes a task.

According to COCOM, the control mode is completely determined by the task environment and has nothing to do with specific task behavior. CREAM summarized the task environment into nine different factors.

The CREAM method summarizes the environmental impact factors into nine major factors, collectively referred to as common performance conditions, namely: organizational integrity, working conditions, human-machine interface and operational support perfection, cooperation of team members, available time, procedure/plan availability, adequacy of training and experience, time zone, and number of simultaneous targets. As the CREAM method is applied to the assembly line, the nine factors cannot be fully applied to the assembly line, so the CPC(Common Performance Condition) factors are adjusted from 9 to 8, as shown in Table 2.

The factors listed in Table 2 are called general performance factors (CPC, Common Performance Condition), and each CPC has three different levels, which can be divided into improved, non-significant, and reduced conditions for human reliability.

The CREAM method uses the CPC level to describe the task environment. The importance of each CPC factor should be different in different task environments. Therefore, we must first determine the corresponding CPC weights, denoted as ${{W}_{j}}$, for different task environments. jcorresponds to Table 2 in the impact of the number of the factor.

The task environment can be quantified by scoring the CPC.

First, we can determine a single score and name it ${{S}_{j}}$.

Specific treatment: the improvement effect corresponds to 1, the reduction effect corresponds to -1, and the non-significant corresponds to 0.

${{S}_{j}}=\left\{ \begin{matrix} 1, \\ 0, \\ -1, \\ \end{matrix}\begin{matrix} \text{Improved} \\ \text{ Not significant } \\ \text{Reduced} \\ \end{matrix} \right.$

In this paper, we use the improvement of the CREAM method by Liao [17]. Obtaining a total score by adding or subtracting the improved and lowered scores has a certain flaw.

Here, we set two variables: G = ∑ to improve the total score andJ = ∑ to reduce the total score.

Divide the task environment factors (CPC) into two broad categories. The first category only considers the CPC with the improvement function in the production environment, and the corresponding MFR (mean value of the failure rate) is recorded as MFR_G. This means that what is positive for human reliability is in this part.

The second type considers only the CPC that plays a role of reduction in the production environment, and the corresponding MFR is denoted as MFR_J. This means that those who have a negative impact on people’s reliability will be in this part.

There is also a basic environment: all the influence of CPC on human reliability is not significant. ${{S}_{j}}=0,\text{ (}j=1,\text{ }2,\cdots ,9).$ The corresponding MFR_G is called MFR₀. The relation between MFR_G, MFR_J, MFR₀, G, andj is as follows:

First, the basic valueMFR₀is determined. According to the CREAM method proposed by Hollande in 1998, the failure probability basic value of 13 types of cognitive dysfunction is given, so the average of the 13-failure probability basic values is 0.0067. Then,k₁ andk₂ are to be determined.

Some basic data from the CREAM method are included in Table 1 and Table 2. Assuming 9 CPC weights are equal, take 1/9. It can be seen from Table 2 that the maximum value of G is 7/9 and the minimum value of J is -1. As shown in Table 2, when Gis taken as 7/9, the smallest value of MFR_G is taken as 0.0001, and when J is -1, MFR_Jis taken as the maximum value of 1.

To take the data into Equations (3) and (4), k₁= -2.35, k₂= -2.17. We can figure out:

Therefore, the prediction model of failure probability of the person can be established as:

Bring Equations (5)and (6) into Equation (7). The final probability of the human failure probability prediction model is:

The importance of CPC factors in the assembly line environment is calculated by using the point-to-point comparison method. In this paper, the weight of the following two comparisons is taken as an example, and the other indicators are followed by analogy. Finally, the cumulative score is written down in Table 3.

In this paper, we take the organizational perfection as an example, and the other indicators are the same. Finally, the cumulative score is written in Table 3.

The information in the table is explained as follows: The first item is compared with other items in turn. If the first item has a greater impact on the human error rate than the other items, then fill in 1. If the other items have a greater impact on the human error rate than the first item, fill in 0. The comparison of the item itself is filled with the number 0.

When the item-by-item comparison is completed, the filled scores are added horizontally. The sum value is written in the last column. The value of the last column is the number of each effect on the human error rate.

After all values are determined, the final values are weighted.

The weight is obtained by Equation (9):

The last column of Table 3 is substituted into Equation(9). Obtain the weight of each CPC, as shown in Table 4.

Equation (1) and Equation (2) are used to calculate the values of G and J respectively, as shown in Table 5.

Bring them into Equation (8):

The probability of human error on the assembly line operating system in the current mission environment is 0.0029. The situation is better.

The LEC evaluation method [18-19] is also known as the risk assessment method for operating conditions (Graham-Kinney method). It is a semi-quantitative evaluation method for operating personnel in a potentially hazardous environment. It is used to evaluate the risk and harm of operating personnel in a potentially hazardous environment.

The method uses the product of three factors related to system risk to evaluate the risk of operator casualties: L (likelihood, probability of accident), E (exposure, the frequency with which personnel are exposed to hazardous environments), and C (consequence, the consequences of an accident).

Different scores are determined for different levels of the three factors, and the product D risk is evaluated by the product D of the three scores (danger).

This method was originally applied to the safety evaluation. Different values of L, E, and C were used to determine whether the final project studied was safe. In this paper, it is properly improved, and it is introduced into the automobile assembly line to calculate the reliability of the production line.

In order to be more suitable for the research object of this paper, we change the meaning of LEC assessment method in this paper as follows: factors influencing the stability of working conditions include L (the possibility of the shutdown), E (the frequency of people exposed to the false operating environment), and C (the possible consequences of a shutdown).

The values of L, E, and Care shown in Tables 6 and 7. The reliability of operating conditions is evaluated with three factor values: The LEC Law is evaluated by the occupational $D\text{ }=\text{ }L\text{ }\times E\text{ }\times C\text{ }$ health, equipment, production, safety technical personnel, and human factors expert group who are familiar with the operating conditions in this article. L, E, and C are graded according to the prescribed standards. The reliability score D is calculated to evaluate the reliability level of the operation. The larger the D value, the higher the influence of factors on the reliability of production line operating conditions.

L indicates the possibility of an accident or a dangerous accident, which reflects the frequency of the accident. In the original LEC method, the value of L is determined on the probability of occurrence of an accident. Such a determination method is very subjective. The probability of an accident means our alertness to danger. Therefore, this article intends to make L’s subjective judgment more objective. Since this paper mainly studies the influence of human factors on the reliability of the production line, the artificial error rate calculated in the previous section is used for the value of L.

In this paper, L mainly considers the occurrence of human error, i.e. the value is 0.0029.

E indicates that people are exposed to the frequency of hazardous environment. For the production line, the time that workers are in a dangerous environment is actually the working hours of workers on the production line. During work, workers face risk factors such as production machines and other workers. For the production line, turning the frequency of exposure to continuous exposure in work line is more suitable, as shown in Table 6.

The amount of time workers work on the production line determines how long they are in danger. For the research object in this article, the working hours of the factory are generally 8:30 am to 12:00 noon and 2:30 to 5:30 in the after noon Occasionally, there may be overtime. At this time, the working time will be greater than the normal time.

C indicates the possible consequences of an accident. In other words, the assignment of C is the probability of the death toll after the accident.Due to the extremely low mortality rate in the actual production process of the production line, it is unreasonable to apply the assignment and meaning of C to the production line. Therefore, C is reassigned. Since it is the judgment of the artificial index on the production line, it is the determination of the state of the situation, so the definition of C reassignment is shown in Table 7.

We redefine C as the time it takes for the production line to stop for some reason. For example, when a worker operates a wrong machine, it may cause a machine malfunction. Then, the production line will be repaired due to the sudden use of the machine.

This situation is very likely to cause the entire production line to stop running because one machine cannot run.The length of machine downtime can indicate whether the line is reliable. The longer the stopping time, the greater the influence of this item on the reliability of the production line.

Select a heavy truck cab assembly line as the research object. The assembly line is a mixed flow production line. According to the actual situation of the assembly line, find the production line as a reliable risk factor. The reliability score is calculated as shown in Table 8.

The human error rate L obtained above is filled in the table, and the values of E and Care obtained from Table 6 and Table 7. Here is a description of the values of E and C in the pathological factors. Since workers on the production line do not get sick frequently, if there are workers who are sick, there will be other workers to replace the min time, so the values of E and C are the last ones.

Due to the high accuracy of the probability value of the human error rate, the calculated D value is not easy to compare. Thus, it will be 10,000 times greater for each of the final D values. The resulting D value is then compared in size. The higher the D value, the greater the influence on the reliability of the production line, and the more attention everyone on the production line needs to pay.

It can be seen from the data that the workload of workers in the production line is the first factor influencing the reliability of the production line. The workers’ heart burden not only is in the production process, but also affects the workers’ mood when the work is completed or when they take rest breaks. The heart burden is greatly affected by the workers’ re-operation. If they have sloppy psychology, fluky psychology, complacency, rebellious attitude, speculative psychology, psychology of the rash, lazy mood, blindly, excitement, or blundering mood, this will affect not only the production efficiency but also the product quality.

Personal judgment and reactions as well as work attitude and proficiency are the influencing factors of the reliability of production lines. When workers find themselves operating improperly, they can make the right responses and judgments without causing dire consequences. Workers can rely on years of experience or pre-training to have a correct judgment of emergencies and timely remediation. This has a significant impact on the operation of the production line and is a major factor affecting the reliable operation of the production line. Of course, the ability to use the correct posture and the ability to be familiar with the work procedures and techniques of the position will also affect the reliable operation of the production line.

Physical load and operation method are the main factors that influence the reliability of the production line. The operators have a small body load, so the body load is not particularly important to affect the reliable operation on the production line. However, if the operator’s body appears to be in a state of overload, his reactions and operations will be delayed, which may seriously affect the production efficiency of the production line. There are not too many cases of workers getting sick, so the pathological factors have the lowest impact on the reliability of the production line.

After summarizing the above paragraphs, we get Table 9.

In view of the above situations, we must do a good job in the heart of the production line workers. For example, we can make a comprehensive understanding of workers’ family situations, psychological stress conditions, mental health, emotions, and personality traits as well as conduct regular surveys and inquiries about workers.

Additionally, ask about the satisfaction with the production environment, living conditions, and salary system. If there is anything unsatisfactory, collect unsatisfactory opinions. Through the judgment and decision-making of the leadership, the actual needs of the workers are solved to meet their psychological and life requirements.

Work attitude and working methods should be properly regulated. Keep abreast of the dissatisfaction, resistance, and negative emotions that workers have in the production process and correct them in time. The project must be regularly trained and have a clear understanding of the operational methods and working environment. Inspect the production line regularly. See if the worker’s operating procedures meet the specified standard operating methods. Warn and record non-conforming operations; in severe cases, fines and other strict measures may be adopted.

If workers are to save themselves in case of danger, they must be trained, not just in theory but in practice. The working attitude and working methods should be rationally regulated. Regular production safety study meetings and activities should be held. Report and learn the unsafe incidents and unqualified operations recorded on the production line. Workers must be trained in production safety. Topics such as how to save yourself in the event of an emergency and how to deal with sudden machine failures need to be learned.

In this paper, the improved LEC method is used to study the production efficiency and stability on the assembly line. We can correct the L of the LEC method to apply the CREAM method to calculate the human error rate. To a certain extent, the result of L becomes objective, and the result can reduce the subjectivity of D.

Meanwhile, two values of E and Care modified for the assembly line. By using the improved LEC method, the reliability score of the reliability of the production line is calculated, which is consistent with the actual statistical results of the production line, so the model has certain reference value.

Of course, this paper only increases the factor L (accident probability) to the artificial error rate. There is no more objective method for the value of the other two factors. This is a direction that deserves more in-depth research.

At the same time, under the conditions of production sites, the human-machine environment has a great impact on the probability of accidents. People and machines, machines and the environment, and the environment and the people will interact with each other. Objectively predicting the probability of line failure under three effects also requires more research.

This work was financially supported by the Natural Science Foundation of Inner Mongolia Municipality (No. 2016MS0536).