Novel Approach to Diagnose Safe Electrical Power Distribution

Abstract

The integrity of the $12\mathrm{V}\mathrm{dc}$ power distribution system on a vehicle is essential to guarantee continuous power supply to safety-relevant consumers. Safety-relevant consumers are critical loads, for example, electric power steering, braking systems with functionalities like Anti-Lock Braking or Electronic Stability Control, and autonomous drive systems. To prevent insufficient power supply for safety-relevant consumers due to an increased wiring harness resistance, a novel diagnostic approach is developed to determine the condition of the power distribution, especially the electrical resistance. The influence of measurement errors and bus commutation on the estimation is investigated by using a simulation. By using the diagnostic, a resistance determination in the milliohm range with a standard deviation of $\sigma =0.3\mathrm{m}\Omega$ can be achieved under realistic conditions. This ensures that failures in the wiring harness can be identified, avoiding cascading effects and minimizing recalls. Compared to the state of the art, redundancies, costs, and weight can be saved with the proposed diagnostic system based on electrical resistance estimation.

1. Introduction

Failure in the operation of safety-relevant systems can cause losses of human lives, regulatory penalties, damaging recalls of vehicles, damage to the brand, and liabilities for vehicle manufacturers. In this context, the power supply of the safety-relevant applications is of crucial importance since, without it, the respective application cannot be performed [1,2]. Both the literature and practice show that faults in the wiring harness occur across different manufacturers and vehicle segments. It is therefore essential to have a diagnostic that can identify faults in the wiring harness with sufficient reliability using existing automotive measurement systems. The integrity of the safe power supply for the safety-relevant systems is provided by the power feed, the freedom of interference in the powernet, and the power distribution. The power distribution always consists of a wiring harness, which is made up of different components such as cables, connector systems, and splices. A typical wiring harness path for the connection of an electric power steering system is shown in the Figure 1 [3].

This figure shows that the electric power steering (EPS), as a typical example for a safety-relevant consumer, is connected to a power distribution module (PDM) on the left-hand side and to the ground on the right-hand side. To ensure power distribution to safety-relevant consumers, either the consumer can be supplied redundantly or a diagnostic must determine that the condition of the wiring harness is in a proper state. Firstly, the functional safety requirements for the wiring harness are discussed. The fault modes of a wiring harness are then examined, whereby an increase in resistance in the components can often be found as a fault in the literature and practice. The diagnostic approach and its implementation are then presented. The approach is then analyzed with the use of simulations, including real measurement deviations, and demonstrated in an experimental study in a vehicle. Finally, the advantages of the approach, additional to achieving functional safety, are highlighted and an outlook is given.

1.1. Safety Requirements for Power Distribution

In the wiring harness, the following three failure modes are relevant and must be considered for safe power distribution [4]. The first failure mode, a short circuit to ground (SCG), is a connection of the positive potential to the ground potential of the vehicle’s chassis. This fault is disconnected by a conventional or electrical fuse to protect the wiring harness against overcurrent. An SCG can be prevented by a full taping of the wiring harness and robust routing through the vehicle. The second failure mode, an open circuit (OC), can occur when a connection breaks off. This can be avoided by measures such as strain relief or fault-tolerant connector systems. In summary, SCG and OC failure modes lead to consumers without power supply and occur immediately. This fault is diagnosed due to the corresponding consumer is no longer operational or is not sending any signals.

The third failure mode, a high resistance (HR), results in excessive voltage drop across the wiring harness and, therefore, in a fault or restricted function of the safety-relevant consumers. Hence, an increase in wiring harness resistance must be taken into account in the design, by defining a worst-case voltage drop over the wiring harness path. Over the life cycle, the resistance increases due to environmental influences such as temperature changes and vehicle vibrations, causing the voltage drop to be above the defined worst case before the end of the vehicle’s life cycle.

The HR failure mode is critical for load-dynamic safety-relevant consumers, when the resistance of the wiring harness path is so high that nominal currents lead to such an voltage drop. Therefore, an HR failure mode does not have impact during normal operation but causes a high voltage drop at high power demands that might occur in critical situations. For example, an EPS can fail during a power-intensive evasive manoeuvrer due to excessive voltage drop across the wiring harness. Without the new diagnostic approach the resistance increase of the wiring harness will be not diagnosed and hence, is a latent fault. This is particularly relevant for safety-relevant consumers that are only supplied via a single wiring harness path and therefore have no redundancy available.

1.2. Increase of the Wiring Harness Resistance over Lifetime

Figure 1 shows the typical wiring harness path from a power distributor to the EPS and from the EPS to the chassis ground. Every component within this path can increase in their resistance [5].

1.2.1. Environmental Stresses on Components

Every component of the wiring harness is exposed to environmental stresses, leading to different failure modes. The main influences are temperature, corrosive gas attack, vibration, humidity, and moisture [6,7]. Each component reacts in a different way to the environmental stresses.

Connector Systems

As shown in Figure 1, three connector systems are necessary. The interaction of the stresses can lead to a change in the resistance, whereby the resistance of the connector system can increase by a multiple of the initial resistance [8,9,10,11]. The most common cause of resistance increase is due to vibrations and temperature fluctuations, which lead to the wear of the contact surface. This wear leads to oxidation. As a result, at high currents, the connector system can be thermally overloaded and may fail [12,13]. The extent to which connector systems are prone to degradation depends on a number of different factors, such as actual current flow, the material of the contact surface, or tightness against water entry. Therefore, field studies were carried out concerning the failure modes of connector systems. As found from [3], in (maximum) 36-month-old vehicles, about 15 percent of the failures are due to a random cause, such as corrosion. For high-mileage vehicles, field tests by [6,14] have demonstrated a resistance increase of more than three times of the initial resistance. In Laporte [15], the contact resistance is rated as faulty when it has increased by $4\mathrm{m}\Omega$. Faults in connector systems can lead to extensive recalls [16].

Bolted Connection System

As shown in Figure 1, the EPS is connected to the ground via a screw connection system. Experimental investigations, issues in the field, and [5] suggest that an increase in resistance can occur on screw connection systems. Factors such as a decrease in the connection force due to the loosening of the bolted connection, temperature fluctuations (especially when using different metals), and wear have an influence on the conductivity [17]. A connection system insufficiently bolted to the ground can lead to recalls in the field [18,19].

Electrical Wiring

As shown in Figure 1, two wires are required to connect to the power distribution module and to connect to ground. Their insulation can be damaged by thermal cycling or external damage. This can lead to the SCG or OC failure mode. However, if a wire is used according to its defined specification, no resistance increase is expected [20].

1.2.2. Misapplication or Improper Design

In addition to the degradation of the wiring harness, the resistance may also increase due to misapplication or improper design. These refer, on the one hand, to the different phases of the life cycle, such as manufacture, assembly, or customer operation, and, on the other hand, to correct use, such as compliance with the maximum permissible operating temperature and the minimum bending radius or protection against mechanical damage. If the defined specification and application of each component are not complied to, these can lead to the failure modes SCG, OC, and HR [3,5,21]. To avoid these failure modes, vehicles at risk of an inadequate connection are recalled for inspection [22].

1.3. Novel Diagnostic Approach

Studies have shown that wiring harness resistance can increase over the service life of a vehicle. In the worst-case scenario, this can lead to recalls and, therefore, to enormous costs for car manufacturers. The safety requirements derived from Section 1.1 and studies by [5] led to the conclusion that a reliable diagnostic to identify an increase in resistance contributes to the safe power supply of safety-relevant consumers. This means that there are multiple factors that require a diagnostic to determine the wiring harness resistance.

To match the demand, the diagnostic must fulfill the following three requirements. First, the diagnostic must reliably determine the electrical resistance of the wiring harness path. In this way, the failure mode HR should be detected. Second, the influence of the communication bus and different sample times must be taken into account, if the electrical quantities are measured at different electronic control units (ECUs). Third, measurement errors must be considered in the diagnostic. Thus, for each measurement point, it must be taken into account that it can have an additive and a multiplicative error as well as measurement noise.

In order to exclude an HR failure mode due to an excessive voltage drop across the wiring harness, a resistance must be measured and compared to a resistance limit. For this purpose, for example, a maximum voltage drop $U_{\mathrm{drop}}=1\mathrm{V}$ is defined over the wiring harness path. This must not be exceeded when the maximum current occurs. In the case of EPS, this can be up to $I_{\max ,\mathrm{EPS}}=120\mathrm{A}$, which leads to a resistance limit of $R_{\mathrm{limit}}=8.34\mathrm{m}\Omega$ [23,24].

2. Methods

To fulfill the requirements, a novel diagnostic approach to determine the resistance of a wire harness path is presented.

2.1. Physical Derivation

As shown in Figure 2, at least the voltage measuring points $U_{\mathrm{PDM}}$ and $U_{\mathrm{EPS}}$, as well as the current measuring point $I_{\mathrm{EPS}}$, are necessary.

$$R_{\mathrm{EPS}}=R_1+R_2={\displaystyle \frac{U_{\mathrm{PDM}}-U_{\mathrm{EPS}}}{I_{\mathrm{EPS}}}}$$

(1)

Measuring points at the power supply (the battery in Figure 2) are not sufficient, as additional resistance ($R_3$ and $r_4$) is present due to further wiring harnesses and the current being distributed to several consumers. Therefore, voltage $U_{\mathrm{PDM}}$ and current $I_{\mathrm{EPS}}$ must be measured at the power distribution module. The power supply, the EPS, and the distributor have separate connections to the chassis ground. However, investigations show that the voltage drop across the chassis itself can be neglected.

In order to fulfill the requirement from Section 1.3, the diagnostic is designed as shown in Figure 3 and the method is explained in the following.

2.2. Estimation of the Wiring Harness Path Resistance

To estimate the resistance $R_{\mathrm{EPS}}$, an estimator for non-linear systems is necessary to compensate for the influence of the measurement deviation. Various estimation methods, such as least square methods and particle filters, were investigated. The best results were achieved with the extended Kalman filter (EKF). Welch [25] describes the EKF principle design and functionality as well as the nomenclature that will be used from this point on. The EKF estimates a resistance so that the voltage at the EPS is equal to the voltage at the PDM minus the voltage drop across the wiring harness.

$$h({\widehat{R}}_{\mathrm{k},\mathrm{EPS}}^{-},0)=U_{\mathrm{k},\mathrm{PDM},\mathrm{S}}-I_{\mathrm{k},\mathrm{EPS},\mathrm{S}}{\widehat{R}}_{\mathrm{k},\mathrm{EPS}}^{-}$$

(2)

The task of the EKF is to estimate the resistance in order to predict the voltage at the EPS.

$$z_{\mathrm{k}}=U_{\mathrm{k},\mathrm{EPS},\mathrm{S}}$$

(3)

Thus, the residual r can be determined, which reflects the deviation between estimated and measured voltage at the EPS.

$$r_{\mathrm{k}}=z_{\mathrm{k}}-h({\widehat{R}}_{\mathrm{k},\mathrm{EPS}}^{-},0)$$

(4)

The first part of the EKF is shown in Figure 3 as I. Time update. In the first step, the resistance ${\widehat{R}}_{\mathrm{k},\mathrm{EPS}}^{-}$

$${\widehat{R}}_{\mathrm{k},\mathrm{EPS}}^{-}={\widehat{R}}_{\mathrm{k}-1,\mathrm{EPS}}$$

(5)

and the error covariance $P_{\mathrm{k}}^{-}$ with a constant process noise covariance Q are predicted.

$$P_{\mathrm{k}}^{-}=P_{\mathrm{k}-1}+Q$$

(6)

The second part of the EKF is shown in Figure 3 as II. Measurement update. This corrects the estimation by the recent measured values. To achieve this, the Kalman gain $K_{\mathrm{k}}$ is calculated.

$$K_{\mathrm{k}}={\displaystyle \frac{P_{\mathrm{k}}^{-}{H}_{\mathrm{k}}^{\mathrm{T}}}{H_{\mathrm{k}}{P}_{\mathrm{k}}^{-}{H}_{\mathrm{k}}^{\mathrm{T}}+G}}$$

(7)

Using this Kalman gain, it is possible to update the estimate ${\widehat{R}}_{\mathrm{k},\mathrm{EPS}}$

$${\widehat{R}}_{\mathrm{k},\mathrm{EPS}}={\widehat{R}}_{\mathrm{k},\mathrm{EPS}}^{-}+K_{\mathrm{k}}(z_{\mathrm{k}}-h({\widehat{R}}_{\mathrm{k},\mathrm{EPS}}^{-},0))$$

(8)

and the error covariance $P_{\mathrm{k}}$ with the measurement.

$$P_{\mathrm{k}}=(I-K_{\mathrm{k}}{H}_{\mathrm{k}})P_{\mathrm{k}}^{-}$$

(9)

Consequently, the resistance ${\widehat{R}}_{\mathrm{k},\mathrm{EPS}}$ is used as an estimation of the resistance $R_{\mathrm{EPS}}$. This resistance $R_{\mathrm{EPS}}$ is then fed back into III. Synchronization in Figure 3. Thus, the accuracy of the resistance estimation improves with each computation step since it leads to a more accurate synchronization. If the error covariance $P_{\mathrm{k}}$ falls below a value of 0.005, a sufficiently reliable estimation is assumed. If this value is exceeded again, the last estimation of the resistance $R_{\mathrm{EPS}}$ is taken into account. This means that the distance between several current pulses is irrelevant and, at the same time, the entire current pulse is used to improve the estimation.

In this way, the first requirement from Section 1.3, the determination of the resistance, can be achieved.

2.3. Synchronization of the Measured Quantities

The measurement signals must be synchronised, otherwise the determined resistance would be incorrectly using the Formula (1). At least one measured variable is transmitted via a communication bus, such as LIN, CAN, or FlexRay. Thereby, a constant signal delay cannot be assumed due to jitter and delay in the communication bus. Therefore, a repeatedly synchronization is mandatory. Since the signal delay is shorter than the sampling time of the communication bus, a synchronization $r_{\mathrm{delay}}$ can be introduced. This describes the ratio of the delay of the measurement signal D to the sample time $t_{\mathrm{s}}$.

$$r_{\mathrm{delay}}={\displaystyle \frac{D}{t_{\mathrm{s}}}}$$

(10)

By estimating the ratio between delay and sample time, the synchronization can be accomplished, which fulfills the second requirement from Section 1.3.

2.4. Estimation of the Epistemic Measurement Errors

However, the estimation of the resistance would be strongly biased by the epistemic measurement errors. Therefore, a repeatedly adapting calibration is necessary to reduce the influence of the measurement errors. The simplified representation of epistemic measurement errors in the following formula contains the multiplicative error a from the true value u and the additive error b, which results in the measured value $\tilde{u}$.

$$\tilde{u}=au+b$$

(11)

In addition to determining the resistance of the wiring harness path, the EKF simultaneously determines the ratio $r_{\mathrm{error}}$ of the measurement error $R_{\mathrm{error}}$ to the measurement signal $R_{\mathrm{signal}}$, which can be traced back to the resistance.

$$r_{\mathrm{error}}={\displaystyle \frac{R_{\mathrm{error}}}{R_{\mathrm{signal}}}}$$

(12)

The ratio of the resistance error, which is due to a measurement error, is estimated. Reducing the influence of the measurement error on the resistance to be determined. In this way, the third requirement from Section 1.3, measurement errors are considered in the diagnostic.

2.5. Excitation and Optimisation

To determine the resistance, a sufficiently large voltage drop across itself is necessary. This can be provided either by a requested excitation or, in the case of the EPS, by a steering manoeuvrer.

To better compensate measurement errors, the optimization of measurement noise covariance G and process noise covariance Q can be fine-tuned. Furthermore, the resistance $R_{\mathrm{EPS}}$ can be estimated accurately faster, depending on the initial parameters $R_{\mathrm{k}-1,\mathrm{EPS}}^{-}$ and $P_{\mathrm{k}-1}$.

3. Results

The presented method is evaluated in the following. Subsequently, the advantages of using the diagnostic are described.

3.1. Simulations

In order to test the proposed diagnostic approach, it was modeled, including the measurement deviation, in the simulation tool MATLAB Simulink Version 18b. The design of the simulation model is according to Figure 2. First, the functionality is examined using a single simulation example. Second, different measurement error combinations are implemented and their influence on the diagnostic result is analyzed.

3.1.1. Single Simulation Example

Figure 4a shows an example input to the diagnostic. The measurement errors and delay shown here are based on the deviations caused by the measurement hardware used in state-of-the-art ECUs and bus systems. Here, an additive error can be observed since the voltage of $U_{\mathrm{PDM}}$ and $U_{\mathrm{EPS}}$ must be nearly the same if no current $I_{\mathrm{EPS}}$ flows. This can occur, for example, due to a ground offset at the voltage measuring points $U_{\mathrm{PDM}}$ to $U_{\mathrm{EPS}}$. Since the voltage $U_{\mathrm{EPS}}$ is measured at the EPS itself, the measured value must be transmitted to the PDM via a communication bus. IT can be seen in Figure 4 that the edges of $U_{\mathrm{EPS}}$ occur slightly delayed after the edges of $I_{\mathrm{EPS}}$.

Figure 4b shows the output of the wiring harness diagnostic (WHD). The determination of the resistance $R_{\mathrm{WHD},\mathrm{EPS}}$ is shown on the top. The true resistance in this simulation model is $R_{\mathrm{true},\mathrm{EPS}}=6\mathrm{m}\Omega$ and the estimated resistance is $R_{\mathrm{WHD},\mathrm{EPS}}=6.04\mathrm{m}\Omega$. Thus, the resistance of the wiring harness path after the current excitation of the EPS could be determined. With this accuracy of resistance determination, the failures shown in Section 1.2 can be recognized. The true delay ratio is $r_{\mathrm{true},\mathrm{delay}}=25\%$ and the diagnostic estimates this to be $r_{\mathrm{WHD},\mathrm{delay}}=23\%$. These values indicate that the delay can be determined properly and has only a minor influence on the accuracy of the diagnostic. This means that the measuring points can be distributed across several ECUs. The ratio of the measurement error in the measurement signal is $r_{\mathrm{true},\mathrm{error}}=46\%$. The diagnostic estimated this ratio as $r_{\mathrm{WHD},\mathrm{error}}=44\%$. The accuracy of the error determination indicated robustness against measurement errors. This means that the measurement hardware that is already installed in state-of-the-art ECUs can be used. Despite the measurement errors and delay, the diagnostic managed to determine the wiring harness path resistance to the EPS. In this case, the diagnostic was able to ensure that the resistance limit out of Section 1.3 is not exceeded. This means that a failure due to an excessive voltage drop across this path can be excluded and the safety requirement can be fulfilled.

3.1.2. Validation of the Diagnostic

In order to investigate the behaviour of the diagnostic with different measurement errors, a Monte Carlo experiment is performed. The measurement errors were derived experimentally and from the specifications of a PDM and EPS. In this way, a wide variety of different combined measurement errors can be investigated and thus the diagnostic can be studied in different scenarios. A Monte Carlo simulation offers the advantage that the robustness of the EKF can be analyzed. Furthermore, by repeating simulations, the influence of measurement noise can be better reflected and various possible combinations of measurement errors can be investigated. For this purpose, a uniform distribution of the individual measurement errors of state-of-the-art ECU hardware is used for each measurement point and each time delay, which corresponds to a bus delay via CAN. Using Formula (11), the input $\tilde{u}$ is calculated with the errors from

Table 1. Measurement errors for each measurement point and delay.

Error	${\mathit{U}}_{\mathbf{PDM}}$	${\mathit{U}}_{\mathbf{EPS}}$	${\mathit{I}}_{\mathbf{EPS}}$	D
Multiplicative error a	$1\pm 0.0375$	$1\pm 0.021$	$1\pm 0.0372$	0
Additive error b	$\pm 140.1\mathrm{mV}$	$\pm 93.6\mathrm{mV}$	$\pm 2100\mathrm{mA}$	$2.5\pm 2.4\mathrm{ms}$

.

For each simulation run, random values of a and b are chosen for each error individually according to their uniform distribution.

Figure 5 shows the distribution of $R_{\mathrm{EPS}}$ after 1000 simulation runs. The deviation between mean value $\mu$ and true value can be attributed to the random selection of the measurement errors. With an increase in the simulation runs, this deviation becomes lower. The variance can be traced back to the measurement errors. In this example, after 1000 runs, a maximum of $\pm 1\mathrm{m}\Omega$ is misestimated. This must be taken into account to ensure that the resistance limit in Section 1.3 is not exceeded. Thus, it must be assumed that, if the resistance was estimated to be above $R_{\mathrm{WHD},\mathrm{limit}}=7.34\mathrm{m}\Omega$, the resistance limit may have been exceeded. If an initial resistance of the wiring harness is known, a resistance change of $4\mathrm{m}\Omega$, as shown in [15], can also be applied.

Figure 6 shows the influence of the multiplicative and additive errors of the different measurement points. The time delay has no relevant influence on the result and is therefore not shown. The additive errors, which are shown in Figure 6a–c, have a little influence on the estimated resistance since these errors are detected by the diagnostic and are included in the error ratio in (12) accordingly.

The multiplicative errors, which are shown in Figure 6d–f, have the main influence on the deviation. Since these measurement errors of the diagnostic appear like a variation of the resistance, these cannot be easily included in the error ratio in (12). Therefore, they have a direct influence on the deviation of the resistance estimation. However, part of the voltage multiplicative error a of $U_{\mathrm{PDM}}$ and $U_{\mathrm{EPS}}$ can also be reasonably interpreted as an additive error, since the voltage is normally around $12.5\mathrm{V}$, making the multiplicative error appear like an additive error. However, a deviation between true and estimated resistance should always be expected since the EKF always assumes a noise in each measurement. The red lines in Figure 6 show the correlation between the estimated resistance and the measurement error determined by simple linear regression. Furthermore, the regression coefficient is shown at the bottom of each plot. Using simple linear regression, 98 percent of the resistance estimation deviation can be attributed. The remaining 2 percent could be attributed to the measurement noise. Additional investigations of the diagnostic showed that the resistance estimations vary depending on the resistance to be determined and the current excitation.

To further improve the results, each multiplicative error can be additionally estimated by a second measurement point. An additional calibration of the voltage measurement point $U_{\mathrm{PDM}}$ would further improve the result. Another possibility would be varying current excitations. The wiring harness resistance changes relatively slowly over its lifetime. Hence, the evaluation of the estimated resistance can be carried out at the end of the driving cycle.

With the measuring points shown in Figure 2, it is not possible to determine which part of the wiring harness has an increased resistance. This would require additional voltage measuring points, which, for example, measure the voltage behind the connector systems in order to monitor individual increases in resistance.

3.1.3. Experimental Validation

In order to investigate the capabilities of the diagnostics in a real environment, it was implemented in a test vehicle. The diagnostics is running on the PDM and the EPS sends the voltage signal to the PDM via CAN.

The measured voltages and current as well as the result of the diagnostics are shown in Figure 7. A steering input when parking has caused the EPS to assist the driver. This caused a power demand and, therefore, a current flow through the wiring harness of the EPS. This resulted in a voltage offset between the PDM and EPS. This was sufficient for the diagnostics to determine the resistance at the end of the current profile. The true resistance was determined before the tests when the vehicle was modified. The cyclical voltage change is due to an alternating set voltage of the DC/DC converter. The values $r_{delay}$ and $r_{error}$ were not included, as these true values in a real test environment are unknown.

3.2. Advantages by Using the Diagnostic

In addition to preventing failures due to increased wiring harness resistance, the diagnostic provides further advantages.

3.2.1. Diagnostic Instead of Redundancy

In order to fulfill the safety requirement of the safety-relevant consumers for safe power distribution, either a robust connection or a redundant power supply must be used. However, redundancy is always associated with additional weight and costs as well as more installation space. The diagnostic coverage makes it possible to make redundancy obsolete [4,26].

3.2.2. Design with Less Expensive Materials

In order to increase the robustness of connector systems in the wiring harness, different materials are available for coating. Gold, silver, and tin are the most common materials for coating in connector systems. Silver, for example, is more robust than tin but less than gold against an increase in resistance [3,7]. Thus, it is possible to use cheaper materials with the help of the diagnostic.

3.2.3. Miniaturization of the Wire Harness Cross-Section

Another advantage can be gained in the design of wiring harnesses. If resistance can be monitored continuously, fewer worst-case assumptions need to be made to guarantee the robustness of the wiring harness over its lifetime. Studies of connector systems used in high-mileage vehicles indicate that only a small proportion of connector systems begin to degrade [6,14]. Laboratory studies also indicate that some connector systems will not increase their resistance [27]. By monitoring the resistance, an increase could be accepted without being safety-critical. Thus, to reduce cost and weight, a smaller cross-section could be used in the design. However, it should be noted that, if the diagnosed wiring harness path degrades, this will be detected by the diagnostic. This makes the fault easy to locate, but a part or the whole path must still be repaired or replaced in a workshop.

3.2.4. Diagnostic of the Entire Power Supply Path

A sensible extension would be to also consider the power supply of the PDM from the battery in Figure 2. Thus, instead of a threshold resistance, the voltage drop over the entire power supply path $U_{\mathrm{drop}}$ could be considered, taking into account expected currents $I_{\mathrm{EPS}}$ and $I_{\mathrm{Batt}}$. This voltage drop can then be compared against a voltage value.

$$U_{\mathrm{drop}}=R_{\mathrm{EPS}}{I}_{\mathrm{EPS}}+(R_3+R_4)I_{\mathrm{Batt}}$$

(13)

Therefore, an individual resistance increase of a wiring harness component can be better compensated before a warning is given [28].

4. Conclusions

Potential faults in wiring harnesses in automotive applications were investigated, whereby an increase in resistance can occur at different parts of the wiring harness. As a safety hazard, it was found that an increase in resistance in the wiring harness is only detected when failures of safety-relevant consumers, such as the electronic power steering, occur. Therefore, a diagnostic was developed using a parameter estimation, which estimates the wiring harness resistance using an extended Kalman filter. Different measurement errors and delay were modeled and their influence on the robustness of the diagnostic was investigated. This diagnostic estimates reliably the resistance in milliohm range with a standard deviation of $\sigma =0.3\mathrm{m}\Omega$. The results shown were achieved with the measurement accuracy of state-of-the-art ECUs, meaning that no additional measurement hardware is required. As the delay caused by the bus communication had no relevant influence on the diagnostic accuracy, the measuring points can, therefore, be distributed across several ECUs. The diagnostic shows what is already possible with state-of-the-art measurement systems and, thus, differs from approaches such as [29,30].

By using diagnostics, failure modes, such as high resistance and the creeping off-circuit of the shown components of the wiring harness, can be identified and prevented. This means that an insufficient power supply to safety-relevant loads due to an excessive voltage drop can be identified, thus preventing a safety hazard. In addition, faults can be diagnosed more accurately, making recalls more selective. Furthermore, large-scale recalls can be prevented by only recalling vehicles with faults in the wiring harness. In addition to fulfilling the safety requirements, it is possible to save a redundant wiring harness as well as to reduce the costs and weight of the wiring. In further studies, the added value is to be quantified in more detail, along with the extent to which the shown diagnostic can provide further advantages with other systems, such as energy or power management.